METACentrum

New: SCP access to shared NFSv4 disk volumes

Ivana Křenková — Mon, 8 Mar 2010 13:14:53 +0100

We support secure access to NFSv4 (/storage) via SCP. SCP can be used for secure file transfer from the user machine to NFSv4 filesystem (/storage) and viceversa, when encryption is required or NFSv4 is not supported by the user machine. SCP uses SSH (Secure Shell) to encrypt the data transfer. In past, we recommended to reserve a cluster node by interactive task and then make the SCP transfer. Now, the SCP transfers the files directly to the NFSv4 server. Advantages: * it is faster (direct transfer to NFSv4 /storage), * it is more comfortable (no need to wait to run the interactive job), * no need to install NFSv4 on the user machine. Direct access to NFSv4 mounted /storage should be used for communication between cluster nodes and data storage. NFSv4 server: depot1.mc.cesnet.cz. SCP example: * To copy from the local machine to /storage: scp FILE login@depot1.mc.cesnet.cz: copies FILE to user home directory (/storage/home/login) * To copy from /storage to the local machine: scp login@depot1.mc.cesnet.cz:FILE1 . copies FILE1 from /storage/home/login to your local machine. MetaCentrum logins are used. There are 2 authentication methods: * Kerberos - if SCP is run from the machine with configured Kerberos (valid Kerberos ticket required) * Password - use your MetaCentrum password. Detailed instructions can be found at http://meta.cesnet.cz/en/docs/storage/nfsv4/ssh-scp.html http://meta.cesnet.cz/en/docs/storage/index.html MetaCentrum Users Support http://meta.cesnet.cz/

New cluster Tarkil

Ivana Křenková — Thu, 4 Mar 2010 19:23:08 +0100

A new cluster Tarkil (tarkil[1-28].cesnet.cz and tarkil.cesnet.cz) was installed and made available in MetaCentrum (as the replacement of the old cluster Skurut). Specification Nodes 1-28: 2x quadcore Intel Xeon X5570 @ 2.93GHz (2x quad core with HT = 16HT cores), memory 24 GB, disk 2x 300GB SAS 15krpm network gigabit Ethernet, Infiniband 4x QDR frontend: 2x quadcore Intel Xeon E5520 (2.26GHz) memory 12GB, disk 8x450GB More details at http://meta.cesnet.cz/cms/opencms/en/resources/hardware.html The cluster Tarkil is available to all MetaCentrum members thru the testing environment (@arien). All nodes are virtualized. There are 2 virtual machines *-1 (standard queues) and *-2 (preempt) available for users. Infiniband could be used with *-1 only. Disks: /scratch 400GB on *-1, 100GB on *-2 virtual machines. Existing user accounts of the old Skurut cluster were migrated automatically, other users are encouraged to request an account via "My Account -- Additional accounts" section at http://meta.cesnet.cz/. MetaCentrum users support http://meta.cesnet.cz

EGI request user groups to work together

Ivana Křenková — Thu, 4 Mar 2010 12:23:40 +0100

European Grid Initiative EGI request user groups linked to major international projects to work together on a programme of the EGI Technical Forum 2010 conference that will be held from 14th to 17th September 2010 in Amsterdam. The letter can be found at http://meta.cesnet.cz/cms/opencms/cs/news/EGI_Technical_Forum10_invitation.pdf If you are interested in cooperation, please contact us as soon as possible via meta@cesnet.cz. MetaCentrum Users Support http://meta.cesnet.cz/

Contribution to the MetaCentrum Annual Report 2009

Ivana Křenková — Wed, 3 Mar 2010 16:31:03 +0100

We're preparing the MetaCentrum Annual Report covering year 2009. We appreciate all contributions sent in last few years. Although we did not publish any summary report, we used articles for reference purposes in national and international project proposals etc. MetaCentrum Annual Report 2009 will be printed and made available electronically at the MetaCentrum website. We have already asked heavy users of MetaCentrum resources for specific reports on their activities. However, we encourage all users to submit a report on their scientific results achieved using MetaCentrum resoruces in 2009. Detailed instructions can be found at http://meta.cesnet.cz/wiki/Pokyny_pro_p%C5%99%C3%ADsp%C4%9Bvek_do_Ro%C4%8Denky_MetaCentra, deadline for the contribution is March 31. MetaCentrum Users Support http://meta.cesnet.cz/

Accounts on cluster Tarkil

Ivana Křenková — Mon, 1 Mar 2010 16:21:47 +0100

During the process of installation of the Tarkil cluster user accounts were created there for the users having their account at the Skurut cluster. Automatic notification on the account creation was sent on Friday. However, as the cluster installation is still in progress, accounts on all cluster nodes may not have been created yet. As soon as all accounts are created, the system will send another notification automatically. The account on the Tarkil cluster is created but not active for users who have not asked for their account extension for this year. It can be activated in the standard way of MetaCentrum accounts extension on https://meta.cesnet.cz/osobni/personal/prodlouzeni.do. In case the account was not created automatically, you can ask for access on https://meta.cesnet.cz/osobni/personal/dalsiucty.do. The cluster Tarkil will be available to all MetaCentrum members in the testing environment. We will inform you about its availability in next few days. MetaCentrum Users Support http://meta.cesnet.cz/

Decommission of cluster Skurut

Ivana Křenková — Tue, 23 Feb 2010 20:12:02 +0100

Following the owner's request, the Skurut cluster will be decommissioned. New job submission is disabled. As soon as jobs are completed, all the nodes will be switched off. The content of the home directory will be moved to home directories /storage/home/LOGIN/skurut_home. A new cluster Tarkil is being installed as the replacement. We will inform you about its availability soon. We are sorry for any inconvevience and we thank you for understanding! MetaCentrum users support http://meta.cesnet.cz

New MetaCentrum users wiki

Ivana Křenková — Fri, 5 Feb 2010 10:54:11 +0100

MetaCentrum users wiki (http://meta.cesnet.cz/wiki) serves for sharing knowledge, tips, manuals, howtos and recommendations among MetaCentrum users. The English version of the wiki contains a list of documents available in English (a portion of the pages is available in Czech only). The "Czech" main page points to all the documentation available, both in Czech and English. Wiki is open to all MetaCentrum users, you are welcome to help us with filing the wiki. User groups can create customized pages here with help for their users. To edit wiki pages please login in the upper right corner (login and password are the same like to MetaCentrum). MetaCentrum Users Support http://meta.cesnet.cz/

Default quota on skirit:/home/[login] set to 2 GB

Ivana Křenková — Tue, 2 Feb 2010 12:14:37 +0100

Default quota setting on the /home filesystem at cluster skirit was increased from 500 MB to 2GB. Checking our filesystem, we have also found the quotas in directory skirit:/home/[login] were missing for some accounts. Due to a limited size of this volume (2,8 TB) we set the quota for all accounts. We would like to kindly ask you to crosscheck the occupied space and the quota. Current quota setting can be found at http://meta.cesnet.cz in section "My account" -> "Filesystems". Higher quota can be requested here if justified. For huge amount of data for calcultations or to be stored we recommend using the volume /storage/home/[your login] instead. The default quota for the volume /storage/home/[login] is 5 TB (compared to 2 GB at skirit:/home). For large data used for curently running jobs we recommend to use fast local storage -- /scratch. Manuals are available online at http://meta.cesnet.cz/en/docs/aplikace/run/index.html, http://meta.cesnet.cz/en/docs/storage/index.html, and a new Wiki http://meta.cesnet.cz/wiki (Czech version only). We are sorry for any inconvevience caused and thank you for understanding! MetaCentrum Users Support http://meta.cesnet.cz/

Decommission of Perian18-68 nodes (PBS property per2)

Ivana Křenková — Fri, 22 Jan 2010 14:48:01 +0100

Following the owner's request, nodes Perian18-68 (PBS property per2) will be decommissioned. New job submission is disabled since last week. After all jobs are completed, the nodes will be switched off. MetaCentrum User Support http://meta.cesnet.cz

New host Ajax

Ivana Křenková — Mon, 18 Jan 2010 15:41:14 +0100

A new host Ajax (ajax.zcu.cz) was installed and made available in MetaCentrum (it is a replacement of the old Ajax 8x Intel Ithanium 2). Specification: Dell PE R710 (2U) 2x Intel Xeon X5570 @ 2.93GHz (2x quad core HT = 16HT cores) 72GB RAM 2x 300GB SAS 15k 3,5"HDD http://meta.cesnet.cz/cms/opencms/en/resources/hardware.html#ajax Existing user accounts of the old Ajax were migrated automatically, other users are encouraged to request an account via "My Account -- Additional accounts" section at http://meta.cesnet.cz/. MetaCentrum users support http://meta.cesnet.cz

Queue backfill

Vlasta Žáková — Mon, 18 Jan 2010 15:19:26 +0100

We have created new queue "backfill" after an arrangement with MetaCentrum users with large amount of submitted jobs. This new queue should be primary used for large amount of jobs lasting up to 24 hours. Queue priority is low; jobs submitted in queue would "fill" empty machines (that is why the queue is proper for one node tasks). Jobs submitted in this queue can be killed in case of need. Queue detail can be found here: http://meta.cesnet.cz/pbsmon2/queue/backfill We plan to restore limits of jobs amount in queue normal due to this new queue. Please contact us with your questions at address meta@cesnet.cz

Queue privileged

Vlasta Žáková — Thu, 14 Jan 2010 23:01:55 +0100

Queue privileged is now available also to users who wrote a tutorial or documentation of an application which they use in MetaCentrum. We want to support active users who wrote their experience with particular application in MetaCentrum. These active users help newbies to solve their problems with utilization of applications. Your texts will be published on MetaCentrum portal. Example of tutorial can be seen here: http://meta.cesnet.cz/cs/docs/aplikace/software/wien2k.html If you have your own application tutorial which can help the other MetaCentrum users, send it please to address meta@cesnet.cz. Your tutorial will be published and you will get access to privileged queue.

cell ruk.cuni.cz downtime

Michal Voců — Wed, 30 Dec 2009 15:49:05 +0100

The AFS file servers for cell ruk.cuni.cz will be temporarily shut down at 20:00 to allow for disk array upgrade. We apologize for any incovenience caused.

scheduler changes

Martin Kuba — Mon, 14 Dec 2009 15:54:53 +0100

The following changes were made in the PBS scheduler, with the goal to make waiting of multi-CPU jobs shorter: - only exact resources are reserved for starving jobs (CPU, memory and exclusivity), not whole machines as before - starving jobs now respect queue priorities, not just job age as before - fair-share is respected in one queue event among starving jobs - starving job is every job older than 1 minute - the scheduler is now 4-5 tmis faster, that will make processing of short jobs faster in the case of larger number of jobs in the system

5th EGEE User Forum

Jiri Chudoba — Mon, 7 Dec 2009 12:41:34 +0100

We’d like to inform you that the deadline of the Call for Abstracts for the 5th EGEE User Forum has been extended! Please note that the new deadline is Monday, December 14. Don't miss the opportunity to participate in one of Europe's largest user-oriented distributed computing events. To submit an abstract, please visit http://indico.cern.ch/conferenceCFA.py?confId=69338 . The programme committee invites abstracts for contributions in one of following general topics: · Planned or on-going scientific work and results obtained using distributed computing technologies, · Data-intensive and data-oriented distributed applications, · Experiences from application porting and deployment of existing software in distributed computing environments, · Software services exploiting and/or extending grid middleware (gLite, ARC, UNICORE etc), · Programming environments, · End-user environments, scientific gateways and portal technologies, · Emerging technologies within the European distributed computing infrastructures (cloud, virtualization etc), National and international activities and collaborations for share and exploitation of e-Infrastructures, · Support services and tools for user communities. Please also note that the REGISTRATION to the 5th EGEE User Forum is open! In order to save money, we strongly encourage you to register as soon as you submit your abstract and generally before 31st of December, to avoid a 25% VAT surplus (applied by the Swedish government from 1st of January 2010)! To register and book your accommodation, please go here: http://egee-uf5.eu-egee.org/index.php?id=689 Jiří Chudoba.

Cluster skurut reinstalation is finished

Martin Kuba — Fri, 27 Nov 2009 13:00:56 +0100

The reinstallation of the skurut cluster, announced on 21.10.2009, is finished. The OS was reinstalled to Debian 5.0, thus the corresponding PBS property was changed from 'sarge' to 'debian50'. The new OS contains NFSvč and thus the /storage directory is available.

power outage at RUK computing facilities

Michal Voců — Fri, 27 Nov 2009 10:10:42 +0100

Due to the unexpected power outage at RUK computing facilities the servers acharon and dali were temporarily shut down. The AFS cell ruk.cuni.cz was also partially inaccessible. All services were back in operation this morning. We apologize for the inconvenience caused.

Power outage

Miloš Mulač — Fri, 20 Nov 2009 12:49:17 +0100

we would like to remind you the power outage on Sunday in Pilsen. Please, consider stopping your jobs to avoid data loss.

Transformers change in Plzeň

Vlasta Žáková — Tue, 10 Nov 2009 09:07:44 +0100

There will be a planned outage of clusters nympha, konos and glamdring on Sunday 22.11. between 10:00-13:00 due to change of measuring transformers.

Outage of cluster skurut

Vlasta Žáková — Thu, 5 Nov 2009 10:45:03 +0100

There will be two short planned outages of the machines skurut1-9 and skurut33-68 on Saturday 14.11. First is planned about 8:00 and the other between 12:00 and 14:00 due to a reconstruction of backup power supply. We are sorry for the inconvenience.

Planned outage of clusters in Brno

Vlasta Žáková — Wed, 4 Nov 2009 09:03:54 +0100

There will be a planned outage of the clusters skirit, loslab, perian and manwe on Saturday 7.11. and Sunday 8.11. We are sorry for the inconvenience.

Clsuter cksurut reinstallation

Martin Kuba — Wed, 21 Oct 2009 17:22:21 +0200

The operating system on all machines in the cluster skurut will be reinstalled to Debian 5.0. Running jobs will be allowed to finish. The machines have insuficient memory, so they will not be virtualized, but after the reinstalltion NFSv4 will be available and thus the /storage directory. The OS change will be reflected by PBS property change from 'sarge' to 'debian50'

changes in the testing environment

Martin Kuba — Tue, 20 Oct 2009 11:46:30 +0200

We are planning changes in the testing environment due to an upcoming deployment of virtual clusters: 1) the clusters Alela and hermes and the machine Manwe7 will be moved from the testing environment to the production environment. i.e. moved from the PBS server arien to the PBS server skirit-f 2) the machines skirit49 up to skirit84 will remain in the testing environment, however there will be a change in their virtualization (the -2 machines will have selectable images), thus they may be more unstable than usual We plan to make the changes as soon as possible, probably this week, but after the currently running jobs finished.

new cluster Eru

Martin Kuba — Mon, 19 Oct 2009 14:54:06 +0200

The MetaCentrum was enlarged with the new cluster Eru consisting of two machines eru1.ruk.cuni.cz and eru2.ruk.cuni.cz. Each of the machines has 8 quad-core processors AMD Opteron 8384 2.7GHz and a relatively large memory - eru1 256GB and eru2 128GB. See http://meta.cesnet.cz/cms/opencms/en/resources/hardware.html#eru The cluster Eru is available to all MetaCentrum members thru the queues short and normalin the testing environment. You can ask for you account on the new cluster on the portal in the section My Account - Additional accounts. When submitting jobs please consider that jobs not needing shared memory will be probably computed faster on several machines with lower number of CPU than on a machine with many CPU.

Grid Computing Seminar 2009

Jan Kmunicek — Mon, 19 Oct 2009 11:29:54 +0200

Hi, we would like to cordially invite you to participate at Grid Computing Seminar 2009 organized by project MetaCentrum and SGI on November 12, 2009 in Brno. The aim of the seminar is to provide overview of project MetaCentrum, describe its current status and especially implementation of new services including future planned ones. Moreover, you will have the unique possibility to discuss your computational needs, plans and potential problems directly with the MetaCentrum operation leader Zdeněk Salvet who will be personally attending the seminar. Detailed information including registration are available at address http://meta.cesnet.cz/en/about/seminars/seminar2/index.html If you know anybody in your vicinity who could benefit from this announcement, please, be so kind and spread this information. Best regards, Jan Kmunicek (MetaCentrum user support)

UPS failure on cluster konos

Miloš Mulač — Thu, 8 Oct 2009 11:39:45 +0200

at 10a.m. the UPS powering cluster konos failed. Nodes konos and konos11-37 went off.

failure of certain AFS volumes

Zdenek Salvet — Wed, 7 Oct 2009 18:40:01 +0200

today between 3pm and 4pm some writable AFS volumes in cell ics.muni.cz (especiallu AFS home directories) have been inaccessible due to hardware problem. We apologize for the inconvenience.

Reboot of konos

Miloš Mulač — Thu, 1 Oct 2009 15:27:21 +0200

frontend of cluster konos had to be rebooted due to error in OS.

Cluster konos failure

Miloš Mulač — Thu, 24 Sep 2009 12:53:04 +0200

due to power shortage at 10a.m. same nodes of cluster konos went off.

circuit breaker change at cluster hermes

Martin Kuba — Fri, 18 Sep 2009 12:21:07 +0200

There will be a planned outage of the cluster hermes on Tuesday 22.9.2009 from 8:00 till 18:00 due to a change of circuit breakers and adding a UPS.

skirit.ics.muni.cz rebooted

Zdenek Salvet — Fri, 18 Sep 2009 12:04:21 +0200

skirit.ics.muni.cz had to be rebooted today around 11:08am due to system panic (OOPS). We are sorry for the inconvenience.

queue preemptible

Martin Kuba — Mon, 3 Aug 2009 17:17:23 +0200

The last news wrongly announced that the cluster orca is accessible by queues short and normal. But that would mean that the jobs submitted thru the orca queue would do preemption on jobs submitted thru the queues short and normal, which would in turn delay them for up to 30 days. This may be unacceptable. So we have created a new queue named "preemptible". It is similar to the queue long, but its jobs can be preempted. Thus the cluster orca is accessible by all MetaCentrum users thru the queue "preemptible".

Clusters perian and orca

Vlasta Žáková — Mon, 3 Aug 2009 13:42:38 +0200

Clusters perian and orca, which were available to users from NCBR, are available to all MetaCentrum users now, throught queues 'short' and 'normal'. The users who are also NCBR members have priority access to cluster throught queue 'ncbr' and 'cpmd' (cluster perian) and 'orca' (cluster orca). Different mode is at the cluster orca. Jobs of J. Šponerová can hold the other jobs at machine orcaX-2. That is why only the single-processor jobs will be allowed at machines orcaX-1.

Air conditioning outage in Pilsen (nympha, konos) on Monday Aug 3rd

Zdenek Salvet — Thu, 30 Jul 2009 11:25:23 +0200

air conditioning system will be out of service in ZCU server room in Pilsen (where clusters nympha and konos are located) on Monday August 3rd since 8:00am. The clusters will be switched off at the time, return to service is expected in the afternoon.

/storage and skirit:/home out of service on Tuesday August 11th

Zdenek Salvet — Tue, 28 Jul 2009 10:49:17 +0200

/storage and home directories of Brno-located clusters will be out of service due to hardware and firmware repairs on Tuesday August 11th since 9:00am. The downtime of approx. 10 hours is to be expected.

sdisk array available again

Ladislav Havlík — Thu, 23 Jul 2009 11:24:08 +0200

Disk array failure we have announced yesterday (/storage, /home of the cluster skirit, perian, loslab and manwe) has been repaired and is fully available again. Best regards Ladislav Havlik

failure storage element

Ladislav Havlík — Wed, 22 Jul 2009 15:12:06 +0200

The storage element has failure now, thus the shared directories /storage, /storage/software, /home are now available until further notice. We prepare to repair the broken disk in the storage element. We apologize for any inconvenience.

hardware upgrade of perian77-perian96

Zdenek Salvet — Mon, 20 Jul 2009 14:51:45 +0200

machines perian77 through perian96 (having PBS property "per4") will be unavailable on Wednesday 22th at 9:00 due to hardware upgrade (additional processors and memory). No new jobs will be started since Jul 21th, 9:00.

job start failures on cluster nympha

Zdenek Salvet — Mon, 20 Jul 2009 13:51:41 +0200

cluster nympha hes been online today between 10am and noon but jobs weren't started correctly at that time due to inaccessibility of /home filesystem caused by hardware failure at nympha node. Long jobs can not be started on cluster nympha for the time being since we expect further service actions on nympha.

Weekend switch-off of cluster konos and nympha

Miloš Mulač — Thu, 16 Jul 2009 12:52:50 +0200

cluster konos and nympha will be witched off at weekend because of air conditioning repairation. Now, PBS doesn't start new jobs on this cluster.

Maintenance of cluster alela

Jan Kmunicek — Wed, 15 Jul 2009 10:54:56 +0200

Cluster alela will be switched off due to problems with cooling. Therefore it does not accept new jobs anymore. Currently running jobs will be left to finish and subsequently the cluster will be switched off. Restoration of the cluster operation can be expected within 14 days at earliest and it will be announced appropriately.

New application programs

Vlasta Žáková — Mon, 22 Jun 2009 18:24:56 +0200

There are new application programs Molden and Grace available in MetaCentrum see: http://meta.cesnet.cz/en/docs/aplikace/software/molden.html http://meta.cesnet.cz/en/docs/aplikace/software/grace.html

Outage of cluster skurut

Vlasta Žáková — Fri, 19 Jun 2009 18:28:38 +0200

It is necessary to consider all-day outage of cluster skurut at 20.6. because of replacement the cabling.

High-level interfaces for jobs management

Jan Kmunicek — Mon, 1 Jun 2009 12:08:56 +0200

System for computational jobs and application programs management Charon Extension Layer (CEL) developed by National Centre for Biomolecular Research and promoted by MetaCentrum has been fully transferred under MetaCentrum administration (http://meta.cesnet.cz/charonwiki). We would like to kindly announce that its further development has been currently suspend due to capacity reasons nevertheless all applications ported under CEL will be in the future fully functional and usable in MetaCentrum environment as well as within the intenational emvironment of virtual organization VOCE. The CEL system itself will be from now on maintained/debugged and extended according to end users needs. Due to the changes in CEL file hierarchy it is necessary to deactivate original CEL and subsequently activate new MetaCentrum CEL installation by following commands: Deactivation of old one (in MetaCentrum environment): /software/ncbr/celcore/meta/migrating/charon-deactivate Activation of new one (in MetaCentrum environment): /software/charon/celcore/meta/migrating/charon-activate Deactivation of old one (in VOCE environment): /software/ncbr/celcore/voce/migrating/charon-deactivate Activation of new one (in VOCE environment): /software/charon/celcore/voce/migrating/charon-activate Moreover, apart from CEL we would like to offer MetaCentrum users to consider other alternatives of high-level interfaces for computational jobs management whose list is available at http://meta.cesnet.cz/en/docs/aplikace/run/frameworks.html. These interfaces are not installed in MetaCentrum, however you are not prevented from a local installation of the specific tool. In the case of any problems or issues, contact authors of the corresponding tool. Best regards and happy Meta computing, Jan Kmunicek

changed ssh keys

Martin Kuba — Fri, 29 May 2009 17:11:37 +0200

Because of the recent security incident, e have changed the host ssh keys on MetaCentrum machines. If you get a warning message when connecting, that begins with @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY! Someone could be eavesdropping on you right now (man-in-the-middle attack)! it is because of the changed keys. A list of the current keys is kept on the machine skirit.ics.muni.cz in the file /etc/ssh/ssh_known_hosts and also on the portal at URL https://meta.cesnet.cz/ssh_known_hosts If you use ssh from Linux or from Cygwin on MS-Windows, you can replace the file ~/.ssh/known_hosts with its content. If you use PuTTy or WinScp on MS-Windows, this information is stored in the registry at keys HKEY_CURRENT_USER->Software->SimonTatham->PuTTY->SshHostKeys HKEY_CURRENT_USER->Software->Martin Prikryl->WinSCP 2->SshHostKeys

back to normal

Martin Kuba — Tue, 26 May 2009 11:52:48 +0200

The planed network outage yesterday in Brno, caused by replacement of UVT's network router, was unfortunately prolonged up to today morning. The expected outage of PBS server and password change service also AFS and thus the MetaCenter portal were unavailable. But now we are back to normal operation. We apologize for any inconvenience.

Actualization of programs Gaussian03 and R

Vlasta Žáková — Sun, 24 May 2009 15:23:05 +0200

Application programs Gaussian03 and R were updated, Gaussian03 to the newest revision E.01 and R to version 2.8.1 see: http://meta.cesnet.cz/en/docs/aplikace/software/gaussian.html http://meta.cesnet.cz/en/docs/aplikace/software/r.html

limited service at Monday 25th evening

Zdenek Salvet — Fri, 22 May 2009 14:44:14 +0200

on Monday 25th May between 19:00 and 20:00 main network router of Institute of Computer Science of Masaryk University will be replaced. Some MetaCentrum services (main PBS server, changing passwords) and some Brno-located machines will be unreachable at the time.

new pbsmon

Martin Kuba — Thu, 21 May 2009 17:25:51 +0200

A new version of the "pbsmon" application is available, which generates the pages on the MetaCentrum portal in the section "Current state". The main changes are: - it distinguishes physical and virtual machines now - it displays mapping of virtual machines to physical ones - the physical machines are arranges in the same way as on the registration form - displays both production and testing environments - there is a new section on the "Machine properties" page which groups machines with same properties - the user detail displays total number of jobs and used CPU time - the machine detail displays queues in which the machine is You can see the new application on the URL http://meta.cesnet.cz/cms/opencms/en/state/nodes If you miss some functions in the application or you have some idea of improvement, we wellcome your suggestions.

cluster move

Martin Kuba — Thu, 21 May 2009 15:02:44 +0200

The alela cluster needs to be temporarily switched off. Thus it is not accepting new jobs now. Running jobs will be let to finish and then the cluster will be switched off.

new cluster alela

Martin Kuba — Tue, 19 May 2009 11:41:15 +0200

The MetaCentrum was enlarged with the new cluster Alela, owned by the Institute of biomedical engineering of Faculty of Electrical Engineering and Communication of the Brno University of Technology. It has 8 nodes, each with 2x Quad Core Intel Xeon E5430 2.66GHz, this total of 64 CPU cores. The memory is 32GB on nodes alela1 and alela2, and 8Gb on the other nodes. The cluster is available to all MetaCentrum members thru the queues short and normal. You can ask for you account on the new cluster on the portal in the section My Account - Additional accounts.

password change needed

Martin Kuba — Mon, 18 May 2009 13:05:38 +0200

There was a security incident on several MetaCentrum machines, and a considerable number of user passwords were stolen by an attacker. Please change your password immediately following the directions on https://meta.cesnet.cz/cms/opencms/en/myaccount/chpasswd If you have logged in from MetaCentrum machines to some other machines, we recommend to change the passwords there as well. As far as we know, this incident was a part of a massive attacks against many institutions at the Czech republic. Thus if you use the same password for other systems, please change it as well. If you use anywhere machines with the UNIX operating system, we recommend to check their integrity. you can use the following commands: strings /proc/`cat /var/run/sshd.pid`/exe /usr/sbin/sshd | \ grep -q msgget && echo "HACKED: sshd modified" ipcs -q | grep -q '07169 ' && echo "HACKED: message queue exists" In the case of a positive outcome, contact your administrator.

skirit.home outage

Martin Kuba — Tue, 12 May 2009 09:52:12 +0200

There was an outage of the shared network filesystem skirit.home, which contains user home directories on clusters skirit,perian,orca,manwe and loslab. The outage was today 12.5. from 2:30 till 9:50. We apologize for any inconvenience.

rapid resources shut-down

Ladislav Havlík — Mon, 11 May 2009 11:35:59 +0200

it is induced reaction for repairing UPS system.

/storage/home available again

Zdenek Salvet — Tue, 5 May 2009 09:43:22 +0200

Filesystem /storage/home is available again. Short outages are expected today due to necessary hardware servicing.

failure of /storage

Martin Kuba — Mon, 4 May 2009 13:50:43 +0200

The NFSv4 server has failure now, thus the shared directory /storage is now available until further notice. We apologize for any inconvenience.

/storage/home fully available again

Zdenek Salvet — Mon, 27 Apr 2009 08:04:02 +0200

/storage/home filesystem has been repaired and is fully available again.

failure of /storage/home filesystem

Zdenek Salvet — Sun, 26 Apr 2009 14:34:24 +0200

serious failure in /storage/home filesystem occured last night. The filesystem is not available at the moment, repair is in progress. We will inform you as soon as the problem is fixed.

Command qsub refining

Martin Kuba — Thu, 23 Apr 2009 15:47:34 +0200

The pages "Personal view" in the section "Current state" were upgraded with a new tool for choosing switches for the command qsub. It displays a set of machines matching a given switch combination, and whether they are free in that time. You can use the tool for checking whether there is enough of required machines, and you can eventually change your requirements to be matched more easily. If there are machines which match your requirements, but you don't have an account on them yet, the tool will suggest such machines to you. The tool is available for both the production and the testing environments: http://meta.cesnet.cz/cms/opencms/en/state/resources/personal http://meta.cesnet.cz/cms/opencms/en/state/testing/personal

New application programs

Vlasta Žáková — Mon, 20 Apr 2009 22:58:42 +0200

There are new application programs available in MetaCentrum: Wine, GCC, Gromacs, deMon and PC GAMESS/Firefly see: http://meta.cesnet.cz/en/docs/software/wine.html http://meta.cesnet.cz/en/docs/software/gcc.html http://meta.cesnet.cz/en/docs/software/gromacs.html http://meta.cesnet.cz/en/docs/software/demon.html http://meta.cesnet.cz/en/docs/software/pcgamess.html

New version of MetaCentrum portal

Vlasta Žáková — Mon, 20 Apr 2009 22:47:05 +0200

Several changes have been currently made in MetaCentrum portal. The sections were systematically split, especially the section "Current state" was heavily edited and completely rearranged. see: http://meta.cesnet.cz/cms/opencms/en/state/resources/nodes

limits on user jobs in queues increased

Martin Kuba — Thu, 16 Apr 2009 09:44:49 +0200

In order to help users with high numbers of jobs requiring few processors, we have changed the limits on the number of jobs for one user on the queue "normal" from 20 to 1000 and on the queue "long" from 40 to 400.

outage of cluster hermes

Martin Kuba — Wed, 8 Apr 2009 15:37:47 +0200

There will be an electric power outage on 25th April 2009 from 8 till 11 o'clock in the building where the cluster Hermes is located. The UPS backups will not be able to cover the whole period, so there will be an outage of the cluster Hermes in that time.

cluster nympha

Martin Kuba — Mon, 30 Mar 2009 10:39:14 +0200

The cluter nympha was moved from the testing environment to the production environment. It consists of 20 machines, each with two quadcore processors, see http://meta.cesnet.cz/cms/opencms/en/resources/hardware.html#nympha

New application programs

Jan Kmunicek — Fri, 20 Feb 2009 11:49:24 +0100

There are new application programs available in MetaCentrum - bioinformatic software MrBayes, QUEEN and X-PLOR and chemical software GAMESS-US. see: http://meta.cesnet.cz/en/docs/software/mrbayes.html http://meta.cesnet.cz/en/docs/software/gamess.html http://meta.cesnet.cz/en/docs/software/xplor.html http://meta.cesnet.cz/en/docs/software/queen.html Best regards and happy Meta computing, Jan Kmunicek

NFSv4 on your personal computer

Martin Kuba — Wed, 18 Feb 2009 16:20:25 +0100

MetaCentrum provides unlimited storage space, which is available on machines marked by the "nfs4" property in PBS in the directory /storage. If you use the Linux operating system on your personal computer, you can use the storage space there as well. We have created a guide how to mount it, it is available at http://meta.cesnet.cz/cms/opencms/en/howto/nfsv4/index.html

outage of portals

Zora Sebestianová — Fri, 13 Feb 2009 11:51:56 +0100

We prepare to move the database and the system Perun, which administrates resources, on a new machine on Monday 16th February cca from 8:00. For that reason all portals will be out of operation.

reconfiguration of cpmd queue and properties of perian cluster nodes

Martin Kuba — Thu, 12 Feb 2009 16:46:33 +0100

According to the utilization analysis of perian cluster nodes with property per2 (e.g. nodes from perian17 to perian68, Xeon, 2.40GHz), the queue cpmd has been reconfigured in the following way: max job duration: 36 hours max number of running jobs: 80 max number of running jobs per user: 50 required property: per2 This change allows users to execute higher amount of shorter jobs on nodes, which are currently less exploited.

liquidation of perian front-end and /perian directory

Martin Kuba — Tue, 10 Feb 2009 11:22:36 +0100

The frontend of the cluster perian will be swicthed off and dismounted, as was agreed with the NCBR administration. That means that also the directory /perian hosted on it will be removed. Please stop using the /perian directory in new jobs. After thirty days even the longest jobs will be finished, so the direcotry will be removed. If any user data remain in it, it will be moved to NFSv4 to /storage/home/login/perian.

cluster minos switch off

Martin Kuba — Thu, 29 Jan 2009 14:21:44 +0100

The machines in the cluster minos will be switched off and recycled on the Monday 2.2.2009. The home directories will be moved to /storage/home/$USER/minos.home and the scracth directories will be moved to /storage/home/odstavene/minosX.scratch where they will be kept for short time and deleted afterwards.

new cluster nympha

Martin Kuba — Wed, 28 Jan 2009 16:59:09 +0100

The cluster nympha was made oprational with new machines today. It now has 20 machines, each with two 4-core CPU Xeon E5472 3 GHz (12 MB cache), 16 GB RAM and 300GB disk. The machines wereadded to the testing environment, thus on the PBS server arien. Users who had account on the old cluster nympha have them on the new cluster.

manwe nad perian failure

Martin Kuba — Fri, 23 Jan 2009 12:39:45 +0100

There was an outage on some nodes of the clusters manwe and perian since 11:00 till 11:30 today (23.1.2009). The outage was caused by a constrution work in adjacent room. We apologize for any inconvenience.

new cluster hermes

Martin Kuba — Wed, 7 Jan 2009 11:52:20 +0100

The MetaCentrum was enlarged with the new cluster hermes, owned by the Faculty of Science of South Bohemia University in České Budějovice. It has ten nodes, each with Quad Core Intel Xeon E5345 2.33GHz and 16GB memory, see http://meta.cesnet.cz/cms/opencms/en/resources/hardware.html#hermes so in total 40 CPU cores. The cluster is available to all MetaCentrum users in queues short and normal. You can ask for an account on the cluster in the portal section My Account - Additional accounts.

Cluster konos service interruption

Miloš Mulač — Thu, 18 Dec 2008 10:25:49 +0100

on Monday 22.12., cluster konos will be temporarily switched off due to UPS upgrade

Repair of /home filesystem

Zdenek Salvet — Tue, 16 Dec 2008 23:44:48 +0100

Failure of disk array skirit:/home announced today has been resolved by moving all data to new disk array attached to different server. Results of running jobs with working directory in /home that have not been stored in the new /home location can be looked up on skirit in /home.old (please don't use /home.old for any other purposes).

Failure of disk array skirit:/home

Zdenek Salvet — Tue, 16 Dec 2008 01:09:33 +0100

disk array carrying /home filesystem shared by clusters skirit, perian, orca, loslab, and manwe has failed on December 15th at 10:24pm. No new jobs will be run on named clusters until repair is complete. Currently running jobs may fail or hang when they access /home. Expected downtime is whole December 16th.

preemption for parallel jobs

Martin Kuba — Fri, 5 Dec 2008 09:54:57 +0100

The virtualization of MetaCenter machines allowed us to provide new feature. In short, it allows special jobs to get ahead of other jobs. We offer this for parallel jobs using more than 10 CPUs. We have created a new queue "preempt" in the testing environment. If you want to use the queue, please contact us ont the email mate@cesnet.cz.

EGEE User Forum

Jiri Chudoba — Wed, 3 Dec 2008 11:18:19 +0100

The 4th EGEE User Forum will take place on 2-6 March 2009 in Catania, Sicily. The event is co-located with OGF25 and OGF-Europe's 2nd international event and will be hosted by INFN. The Program Committee invites abstracts for contributions in one of following general topics: * Scientific results obtained using grid technology * Planned or on-going scientific work using the grid * Experiences from application porting and deployment * Grid Services exploiting and extending gLite middleware (job management, data management, monitoring, workflows etc) * Programming environments * End-user environments and portal technologies * Emerging Technologies within the EGEE infrastructure (cloud, virtualization etc) Contributions may be offered either for oral presentation, posters or demonstrations. All submitted abstracts should follow the pre-defined template provided online from the User Forum web site at CERN's Indico system. Abstracts must be submitted using the "Submit a new abstract page" under the Call for Abstracts section. Prospective authors should read the "Author Instructions" section which provides important information regarding abstract submission. Abstracts submitted via other means (email, etc.) or after the submission deadline will not be taken into consideration. Any enquiries regarding abstract submission should be forwarded to the Programme Committee chair. The full text of the Call for Abstracts is available on-line at http://indico.cern.ch/conferenceCFA.py?confId=40435 Tutorials and Trainings In case you are interested to organise a tutorial or meeting in conjunction with the User Forum, you are invited to contact the Programme and Local Organisation Committee chairs. In case you are interested to organise a tutorial or meeting in conjunction with the User Forum, you are invited to contact the Programme and Local Organisation Committee chairs. However, no meetings and tutorials will be scheduled in parallel with the general event programme. Important Dates Deadline for abstract submission: 5th of December Notification of acceptance: 15th of January PROGRAMME COMMITTEE CHAIR Vangelis Floros, GRNET (efloros@grnet.gr) LOCAL ORGANISING COMMITTEE CONTACT Roberto Barbera, INFN (roberto.barbera@ct.infn.it)

MetaCentrum user database maintenance

Aleš Křenek — Thu, 27 Nov 2008 16:09:04 +0100

Large scale maintenance and migration to new hardware and software version will be done on Monday December 1st. During the maintenance new registrations and requests for accounts will not be accepted and processed, and related functionality of the www portal will be affected (editing of personal information, personalized display of current resource status etc.). On the contrary, running jobs as well as job submission will not be affected. We assume recovering full functionality later in the afternoon. We are sorry for the inconvenience. -- Ales Krenek UVT MU

cluster nympha switch off

Martin Kuba — Wed, 19 Nov 2008 11:43:04 +0100

The machines in the cluster nympha will be switched off and recycled on the Monday 24.11.2008. The home directories will be moved to /storage/home/$USER/nympha.home and the scracth directories will be moved to /storage/home/odstavene/nymphaX.scratch where they will be kept until the 8th December and deleted afterwards. The cluster minos will remain in service so far.

nympha and minos - preparation for HW replacement

Martin Kuba — Fri, 14 Nov 2008 16:18:40 +0100

The machines in the nympha and minos cluster wiil be replaced by new machines before the end of this year. That is why we removed the properties "long" and "iti" from them, so that it is not possible to submit jobs taking up to one month. We will inform you about the replacement in advance.

Outage of disc array at machine skirit

Jan Kmunicek — Mon, 10 Nov 2008 15:42:19 +0100

There was an unexpected outage of disc array at machine skirit during weekend and today. This caused non-accessibility of disc capacities as well as METACentrum portal. We would like to apologize for any inconvenience caused! Best regards, Jan Kmunicek

deleting data from eru and grond

Martin Kuba — Fri, 31 Oct 2008 11:41:52 +0100

Three years ago the big machines named eru and grond were retired. The data from their disks were moved to /afs/ics.muni.cz/odstavene. We think that after such long time the data are not needed anymnore, so we are going to delete them. If you still need the data, contact us immediatelly on the address meta@cesnet.cz.

Planned NFSv4 server smaug1 outage.

Lukas Hejtmanek — Thu, 23 Oct 2008 14:58:41 +0200

The NFSv4 server smaug1 will be in maintenance tomorrow (Friday October 24th) between 14:00 and 14:30 due to HW upgrade.

Grid Computing Seminar

Jan Kmunicek — Wed, 22 Oct 2008 12:03:15 +0200

We would like to cordially invitate you to participate at Grid Computing Seminar organized by project MetaCentrum and IBM on November 26, 2008 in Prague. The aim of the seminar is to provide overview of project MetaCentrum, describe its current status and especially implementation of new services including future, planned ones. Detailed information including registration are available at address http://meta.cesnet.cz/en/about/seminar/index.html If you know anybody in your vicinity who could benefit from this announcement, please, be so kind and spread this information.

planned downtime of Brno-located METACentrum machines on Oct 27

Zdenek Salvet — Wed, 22 Oct 2008 11:13:27 +0200

Some Brno-located METACentrum machines will be shut down on October 27th approx. from 7:15 to 8:15 AM due to repair of UPS. New jobs will not be run there since sunday. Running jobs will not be killed but risk of unplanned crash (power fault) will be higher than usual.

unplanned downtime of Brno-located METACentrum clusters

Zdenek Salvet — Wed, 22 Oct 2008 09:47:49 +0200

Machines were rebooted and running jobs terminated on October 22 at approx. 7:15 am due to finishing of necessary repairs of power distribution equipment. We apologize for any incovenience.

outage of machines in Brno delayed

Martin Kuba — Mon, 20 Oct 2008 09:27:04 +0200

The announced outage of Brno machines was delayed untill tomorrow (21.10.2008)

outage of machines in Brno

Martin Kuba — Fri, 17 Oct 2008 14:32:52 +0200

There will an outage at the computing room in Brno on Monday 20.10.2008 from 7:15 to 7:30. Thus submitting new jobs to machines located at Brno will be blocked from this Sunday.

privileged access in exchange for publications

Martin Kuba — Fri, 17 Oct 2008 13:59:38 +0200

We have introduced the promised privileged access to METACentrum resources. A new queue named 'privileged' was created http://meta.cesnet.cz/pbsmon/queue.do?queue=privileged with priority higher than queues normal and long, with time limit 30 days. The queue is accessible to users with the most publications with acknowledgement ot METACentrum, listed at the portal at https://meta.cesnet.cz/cms/opencms/en/myaccount/pubs The precise way how the users will be selected will be refined over time, now we just coount publication. Later we may count newer publications more. The machines manwe5 and manwe6 were moved from the testing environment to the production environment, so that they ware now in the privileged queue.

NFSv4 at cluster konos

Martin Kuba — Wed, 15 Oct 2008 16:57:47 +0200

The cluster konos has NFSv4 available now. So the 60TB disk array mounted at /storage is available the same way as on the clusters skirit, orca, perian. Also * OS on the frontedn was upgraded to Debian 4.0 to match the cluster nodes * the array mounted to /project was moved to /data-kky/home (concerns users from the group kky) (If you want to receive this news in Czech, change you language preference on the portal.)

NFSv4 on skirit.ics.muni.cz

Martin Kuba — Tue, 14 Oct 2008 13:22:36 +0200

The OS upgrade on the machine skirit.ics.muni.cz was finished today. The main change from the user's point of view is that the 60TB disk array mounted over NFSv4 is now available also from this machine. Each user has a directory /storage/home/ on it. The space on the array is virutally limitless, we plan an upgrade to 100TB in the near future. Just to limit an invalid process from filling up the whole capacity, there is a 5TB quota per user. (Do not be confused by the output of the "quota" command on skirit, it just cannot display such big capacity correctly.) If you need bigger quota, just ask for its extension on the portal at https://meta.cesnet.cz/cms/opencms/en/myaccount/quota You can also see on which machines is the array mounted on the page.

Reinstallation of frontend of cluster konos

Miloš Mulač — Mon, 13 Oct 2008 13:23:01 +0200

reinstallation of OS of frontend of cluster konos will start on Wednesday, 15.10. During the reinstallation directories /home/* and /project/* will be unaccessible on the workernodes of the cluster.

skirit upgrade

Martin Kuba — Thu, 9 Oct 2008 13:46:21 +0200

On Monday 13th October, the OS onthe machine skirit.ics.muni.cz will be upgraded. The machine acts a s the fileserver for clusters skirit, manwe, orca, perian and quark, and also is used as a frontend by many users. In the time of the upgrade the machine and the files hosted on it will not be available.

UPS change in Brno

Martin Kuba — Thu, 9 Oct 2008 13:35:15 +0200

The UPS located at the computing room in Brno will be replaced on Monday 13th October, so there will be increased risk of outage of the METACentrum's machines located in Brno during the time.

Kernel upgrade on skurut

Miloš Mulač — Wed, 17 Sep 2008 12:59:28 +0200

frontend skurut will be switched of terporarily due to kernel upgrade. We plan to start upgrade after 13:30 pm today. During this operation, home directories wil not be available on cluster skurut workernodes.

Short-term outage of manwe3 tomorrow

Jan Kmunicek — Thu, 11 Sep 2008 11:32:12 +0200

There will be a short-term outage of machine manwe3 on Friday 12, September 2008 due to required revision of electrical circuits. We are sorry for any inconveniences caused. Best regards and thanks for understanding, Jan Kmunicek

Clusters switching to a new UPS

Jan Kmunicek — Tue, 9 Sep 2008 13:03:51 +0200

Today there will be switching of selected clusters (skirit, perian, orca) to a new UPS. Only completely free nodes without any user jobs will be switched off and subsequently on. In the case of any problems, do not hesitate to contact us. All the best, Jan Kmunicek

machines not on UPS

Martin Kuba — Wed, 3 Sep 2008 14:58:09 +0200

There was a failure of one of the two big UPS providing electric power to the computing room in Brno on the Friday 29.8. The UPS is realy big, so its repair will take several weeks and it is not possible to get a replacement for it. That is why some of our machines are now running without UPS. Here is the list of the machines: perian 35-37,61-64,83-88 manwe 3 skirit 1-84 aule quark 6,8-13 orca 1,13-181 The machines are up and running, they are just not backed up by an UPS.

Short-term outage of skirit

Jan Kmunicek — Wed, 3 Sep 2008 13:41:36 +0200

There was a short-term outage of machine skirit today at around 1 p.m. (frontend of skirit cluster) most probably caused by hardware failury of disc. Actually, the machine is up and running again. We are sorry for any inconvenience caused. Best regards and thanks for understanding, Jan Kmuníček

outage of Brno node

Zora Sebestianová — Fri, 29 Aug 2008 12:30:24 +0200

By reason of electricity failure and UPS failure all machines on Brno center were down. Several of them are working again on other we are working. Please contact us when some errors occured on working machines. Thank you, service and US group of METACenter

Unplanned downtime of cluster skurut

Zdenek Salvet — Thu, 14 Aug 2008 16:50:25 +0200

cluster skurut had to be powered off due to failure of cooling system of CESNET computer room. We have not estimation of fix time yet.

news and outages in your preferred language

Martin Kuba — Mon, 11 Aug 2008 15:09:59 +0200

The news and outages were so far announced by emails which contained both Czech and English texts. From now on, the emails will contain only one language version, which is selected by the setting of your preferred language in My Account - Personal settings.

OS upgrade on skirits

Martin Kuba — Mon, 11 Aug 2008 13:32:25 +0200

This week (11.-15.8.) is scheduled operating system upgrade on machines skirit numbers 1 up to 48. The frontend machine skirit.ics.muni.cz will be upgraded as well on Wednesday 13.8. The machines will be unavailable during the upgrade. The upgrade will allow NFSv4 filesystem and thus the 60TB disk array mounted on /storage/home, which is so far available on machines orca, perian, manwe[5-7] and skirit[49-84]. We apologize for any incovenience.

InfiniBand

Vlasta Žáková — Thu, 7 Aug 2008 15:40:36 +0200

There is a new high-speed and low-latency network connection InfiniBand available in special test machines. Actual list of machines with InfiniBand property you can find at http://meta.cesnet.cz/testovaci/props.do#infiniband More information about paralell aplications and InfiniBand you can find at http://meta.cesnet.cz/en/docs/software/parallel-applications/index.html

Upgrade OS

Miloš Mulač — Mon, 4 Aug 2008 16:42:28 +0200

OS will be upgraded next week on cluster hydra.

Operation interuption

Miloš Mulač — Mon, 4 Aug 2008 12:06:43 +0200

On Monday evening, 11.8. , cluster minos, konos and nympha will be switched down due to recontruction of power installation. Clusters will start operate on Tuesday, however, prophylaxis of all computer systems will run during rest of the week and may influence the operation of Pilsen clusters.

AFS server on skirit outage

Martin Kuba — Tue, 29 Jul 2008 17:38:46 +0200

There was an outage of the AFS server on the machine skirit.ics.muni.cz today from 14:39 till 16:36. Jobs using files from the AFS server without copies on other Afs servers may be affected. Following directories have no copies: /software/CSRosetta-mar08 /software/RCI /software/adf2007.01b /software/charonGUI-1.0 /software/condor-7.0 /software/dvmsifex /software/etics /software/ganglia-3.0.7 /software/gcc-4.2.3 /software/gmp-4.2.2 /software/ikiwiki-2.41 /software/lb-transport /software/loslab /software/matlab-7.5 /software/matlab-7.6 /software/mpfr-2.3.1 /software/ncbr /software/nfs-utils-1.1.2 /software/nmrpipe-3.0 /software/ofed-1.3 /software/oracle-7.3 /software/remctl-2.12 /software/rosetta-2.3.0 /software/smartmt-5.37 /software/wien2k-08 /software/xman-1.0 /software/xplor-nih-2.20 We apologize for any inconvenience.

Power outage at CESNET

Zora Sebestianová — Mon, 21 Jul 2008 10:58:13 +0200

There was unexpected power outage at CESNET (Prague) on 2008-07-19. Problems related with this power outage are actually resolved (skuruts, shelob and meta-portal etc.). Please inform US if you have problems. Zora

New version of WIEN2k software package

Zdenek Salvet — Thu, 17 Jul 2008 12:20:33 +0200

The newest version of WIEN2k package - 08.2 - has been installed, is can be made accessible using "wien2k-08" module (module add wien2k-08). Four flavors of compiled programs are available, adding wien2k-08 module selects an appropriate one according to machine type automatically. 64bit versions (amd64/em64t, Itanium) have been compiled with parameters NMATMAX=100000 and NUME=10000, 32bit versions (xeon, p3/athlon) with parameters NMATMAX=20000 and NUME=2000.

New 16-CPUs machines

Jan Kmunicek — Thu, 17 Jul 2008 10:28:45 +0200

METACentrum hardware portfolio has been currently extended by two new machines of the same type as cluster manwe - these are concretely 16-CPUs machines manwe6 and manwe7 from which manwe6 has available 64 GB and manwe7 128 GB of memory. Both machines are available in testing cluster through test PBS (http://meta.cesnet.cz/cs/state/testing/index.html). With best wishes of happy META computing, Jan Kmunicek

Results of survey

Vlasta Žáková — Sat, 12 Jul 2008 15:26:28 +0200

The results of survey that has been performed among METACentrum users during last several weeks are published.

Results of survey

Vlasta Žáková — Sat, 12 Jul 2008 14:27:09 +0200

The results of survey that has been performed among METACentrum users during last several weeks are published.

parameter mem change

Martin Kuba — Thu, 26 Jun 2008 13:14:04 +0200

The news from 27th March 2008 announced the activation of the parameter "mem" used for reservation of physical memory on computing nodes. We have found and fixed a bug in PBS regarding the useage of the parameter "mem". A consequence of this fix is that the value of the parameter is no longer applied to a processor, but to a node. So for example when you specify qsub -l nodes=2:ppn=8,mem=1gb each machine will have 1GB of memory reserved, not 8GB. The documentation on portal was updated to reflect the current state.

power outage in Plzeň

Martin Kuba — Thu, 26 Jun 2008 12:27:14 +0200

There was a power outage in Plzeň today between 7 and 11 a.m., and the available backup sources could not provide electric power for so long time. METACentrum machines located in Plzeň thus has experienced the power outage too. We apologize for any inconvenience.

connection outage in Plzeň

Martin Kuba — Mon, 23 Jun 2008 10:22:11 +0200

A backbone network router failure in Plzeň caused today between 8 a.m. and 9 a.m. lost connectivity between METACentrum machines located in Plzeň and the rest of the METACentrum. We apologize for any incovenience.

outage of /home directory

Martin Kuba — Thu, 12 Jun 2008 15:31:22 +0200

A transfer of the physical location of the /home directory on the machine skirit and on all machines that share it, i.e. computing nodes of clusters skirit, perian, manwe, quark and orca, will cause a n outage of it on Monday 16th June 2008 afternoon. It means a short outage of the machine skirit and mainly that jobs running on the mentioned nodes will not be able to use files located in user's home directory. We apologize for any inconvenience.

60TB disk array

Martin Kuba — Wed, 11 Jun 2008 12:54:58 +0200

The storage capacity of METACentrum was just enlarged by a new disk array with 60TB of space. Each user can use up to 5TB of disk space. The disk array is accessible over network file system NFSv4 on machines marked by the property "nfs4" in the PBS. They up-to-date list can be obtained from the production and testing PBS monitoring tools at http://meta.cesnet.cz/pbsmon/props.do#nfs4 http://meta.cesnet.cz/testovaci/props.do#nfs4 In the moment of writing the machines are orca1-* up to orca18-*, perian77-* up to perian96-*, skirit49-* up to skirit84-*, and manwe5. The disk array is physically connected to the machine smaug1, is is mounted on computing nodes to the directory /storage/home, which contains user's subdirectories. You can view the details about your own directories at the page https://meta.cesnet.cz/cms/opencms/en/myaccount/quota where you can also ask for a change of your quota. The NFSv4 network file systems allows to mount the array remotely to your desktop computer, in similar way as the AFS system offered so far, but is much faster (we achieve up to 100MB/s locally). Security is ensured by the Kerberos system.

failure of skirit69-79

Martin Kuba — Mon, 9 Jun 2008 14:30:20 +0200

There was a failure today morning on two surge guards that were protecting physical machines skirit number 69 up to 79. That caused failures of the virtual machines located there (skirit69-1, skirit69-2, ..., skirit79-1, skirit79-2) and also jobs scheduled there. The machines were reconnected to new surge guards and are back in production as of now. We apologize for the job failures and problems that they have caused.

Short-term outage of skirit machine

Jan Kmunicek — Wed, 4 Jun 2008 15:49:49 +0200

Currently there was a short-term outage of skirit machine due to the exchange of disc in disc array connected to skirit. We apologize for any inconveniences caused.

Power outage in Pilsen

Miloš Mulač — Fri, 23 May 2008 16:11:07 +0200

due to power outage, all clusters on Pilsen server room are temporarily out of order.

My Jobs

Martin Kuba — Thu, 22 May 2008 13:08:32 +0200

A new item "My Jobs" was added to the menu "My Account" on the portal. The item displays historical records about finished jobs of the authenticated user. The records come from the planning system PBS and kernel accounting from computing nodes. The direct link is http://meta.cesnet.cz/cms/opencms/en/myaccount/acct The records about jobs from PBS are available since the year 2005, the records about processes are collected during a few recent months. Only records about processes lasting longer than 1 second are collected. The data are collected once daily (after rotating PBS and kernel logs), so it can take up to a whole day after a job finishes before it appears in the list. If you need data about jobs that ended less the one day ago, you can view them in the PBS status, where they are in the state C-completed.

Program ADF in METACentrum

Jan Kmunicek — Tue, 20 May 2008 14:51:07 +0200

Based on your requests a full trial version of program ADF (http://www.scm.com) has been recently installed in METACentrum. Access to this full version is granted after confirmation of licensing policy and utilization conditions. These rules are available at METACentrum portal at address https://meta.cesnet.cz/osobni/personal/eula.do. Therefore, you are kindly requested to confirm your agreement with the conditions. By this step you will be added to the group of ADF program testers and the application module will be made accessible for you. The deadline for ADF testing is June 25, 2008, so do not hesitate to confirm the rules and start testing. The program utilization should be standard, i.e. at amd64/em64t machines (manwe, newest skirit nodes) a modul adf2007 has been prepared. Simple tests produce expected results: module add adf2007 cp /software/adf2007.01b/examples/adf/GO_H2O/* ~/test cd ~/test NSCM=4 ./GO_H2O.run > output Module adf2007 sets several basic variables described in details at http://www.scm.com/Doc/Doc2007.01/Install/Installation/page25.html It is necessary to se the variable NSCM according to number of CPUs that the calculation will use; similarly SCM_MACHINEFILE should be used for definition of machines (when you will use version utilizing the network). These variables have to be set for every calculation separately, in the best way using information from PBS: SCM_MACHINEFILE=$PBS_NODEFILE #for jobs at one machine is not necessary NSCM=`wc -l $PBS_NODEFILE` Documentation is available at http://www.scm.com/Doc/Doc2007.01/Welcome.html, example calculation with optimization of water molecule geometry can be found at address http://www.scm.com/Doc/Doc2007.01/ADF/Examples/page30.html The main reason for making the trial version of ADF accessible for limited time is to find out if the program should be purchased. As ADF is not a cheap product it is necessary to verify its suitability for your purposes; there should be a discussion during end of the trial period concerning next steps especially about potential purchase of ADF. Therefore it is crucial to utilize the testing period to give us concrete experience for upcoming decision. In the case of any problems or questions concerning ADF program (or anything else in METACentrum) do not hesitate to contact us using the usual way through address meta@cesnet.cz. Best regards and happy testing, Jan Kmunicek (METACentrum user support)

Reinstallation of cluster konos

Miloš Mulač — Mon, 12 May 2008 17:02:13 +0200

a new OS (Debian 4.0) was installed on konos37, running two virtual domains konos37-1 and konos37-2. There should be minimal changes from the user point of view, however, versions of some system libraries may differ. We recommend to specify OS version during job submition using node properties 'suse' or 'debian40' for original instalation or for the new one respectively. The rest of nodes will be reinstalled after a short testing continuously. Please, report any problems to meta@cesnet.cz

testing environment

Martin Kuba — Mon, 5 May 2008 17:38:09 +0200

We want to continuously improve the METACentrum environment, but in the same time we do not want to compromise the stability of the computing environment to which our users are used to. That is why we have created a separated testing environment, where we will test new features before they are deployed to the production environment. The testing environment is open to all METACentrum users who want to try the tested features. We added computing resources with interesting power to entice users to become our "voluntary testers". In this moment the testing environment contains the machine manwe5 with 16CPU and 64GB memory, and in the near future it will contains also the machines skirit 49 - 84, which have Infiniband network cards. How to use the testing environment is described at http://meta.cesnet.cz/en/state/testing/ and the state of machines and jobs is at http://meta.cesnet.cz/cms/opencms/en/state/testing/nodes

database connection

Zora Sebestianová — Tue, 22 Apr 2008 13:16:41 +0200

From 11:20 till 12:30 database connection was broken. We are apologized for problems.

privileged access and publications

Martin Kuba — Mon, 21 Apr 2008 17:36:34 +0200

The introduction of discussion forum to our portal was motivated by the need to discuss the planned privileged access to the most demanded machines for those users, who produce the most publications with acknowledgements to the METACentrum. There is already a new section in the My Account portal section, where you can enter your publications. But before we use the data, we want to give you the chance to say your word. Please add your opinion to the discussion at https://meta.cesnet.cz/metaforum/viewtopic.php?f=28&t=5

Discusion forum started

Martin Kuba — Mon, 21 Apr 2008 16:14:13 +0200

We have added a discussion forum to the METACentrum portal, where all METACentrum users can discuss. Links to the forum were added to the main portal page and to the navigation menu, or you can go directly to https://meta.cesnet.cz/metaforum/ We hope that this forum will be the part of the portal, where the content can be added by the users themselves. You can share your experience there, or ask for an advice from the more experienced users. For reporting bugs or problems please continue to use the email address meta@cesnet.cz, which is connected to the RT (Request Tracking) system ensuring that problems are solved.

Unscheduled restart of server skirit.ics.muni.cz

Zdenek Salvet — Thu, 27 Mar 2008 20:10:29 +0100

we had to reboot server due to failure in operating system kernel. Open interactive sessions were closed, jobs using shared filesystem /home mounted from skirit (on clusters skirit, perian, manwe, and aule) could have been influenced too.

parameter "mem" activation

Martin Kuba — Thu, 27 Mar 2008 12:30:20 +0100

Today is the day of activation of checking of the parameter "mem" in the PBS resource planning system. The parameter specifies the amount of memory reserved for a job and, in the same time, the maximal amount of memory a job can use per one CPU. Please start specifying the parameter when submitting a job now. If you don't specify it, the default value is 400MB (which is roughly one half of physical memory in smaller machines with two CPU, which is why it was chosen). When the sum of memory consumption of all processes belonging to a job exceeds the reserved amount of memory on a machine (the value of "mem" times the number or reserved CPUs), the whole job will be canceled. But don't specify the value of "mem" unnecessarily high, it would limit the number of machines where the job can be run, thus prolonging the waiting time before it can be started. Example usage: qsub -q normal -l nodes=1:ppn=4,mem=2000mb job.sh starts the job on a machine with at least 4 CPU and at least 8GB of memory, where 8GB of memory will be reserved for the job and in the same time, all job's processes together must not use more than 8GB of memory. If you have any problems when using the parameter "mem", don't hesitate to contact us at the email address "meta@cesnet.cz".

unplanned skurut outage

Martin Kuba — Wed, 26 Mar 2008 09:09:18 +0100

there was an unplanned outage of the cluster skurut on the night from 25th to 26th March, most likely due to some works at the room where it is located. It is back at service as of now.

memory reservation

Martin Kuba — Thu, 20 Mar 2008 13:54:56 +0100

following users' requests, we will activate more checking of memory reservation by the parameter "mem" in PBS. It will allow better planning of jobs to machines with enough memory, and it will block jobs from consuming memory of other jobs. However it means that all users need to start using the "mem" parameter from that day, as its default value is low. The specified value should reflect the real needs of the job, as a value too big would limit the number of machines where the job can be assigned. More information about resource reservation can be obtained by the command "man pbs_resources". This is an early warning, we will send another notification on the day ot the activation.

Memory upgrade of some nodes of cluster konos

Miloš Mulač — Wed, 19 Mar 2008 10:28:59 +0100

the system memory in nodes konos15 - konos23 was upgraded from 4GB to 8GB. If you want to reservere node with more memory for your job, use parameter 'mem' in qsub command. For more information, see qsub man pages or documentation at http://meta.cesnet.cz/cms/opencms/cs/docs/software/system/pbs_long.html#omezenipameti

retirement of narsil and gandalf

Zora Sebestianová — Tue, 18 Mar 2008 15:40:26 +0100

Machines narsil and gandalf were retired. They were outdated machines with processors Itanium II 1GHz and MIPS 180 MHz. Your accounts at these machines will be deleted.

Failure of part of cluster konos

Miloš Mulač — Thu, 28 Feb 2008 08:54:11 +0100

servers konos and konos11-37 were hit by power cutoff yesterday evening.

CFP CESNET 2008

Martin Kuba — Wed, 27 Feb 2008 09:16:35 +0100

The bi-annual CESNET 2008 conference will be organized on 25th and 26th September 2008 in Prague. It will focus on advances in high-speed communication technologies, grids, security, middleware and virtualisation. For details see the URL. The Call for papers was announced now, the deadline is 15th May 2008. We think that you, as a METACentrum user, may be interested in this conference, so we cordially invite you.

Cluster hydra shutdown

Miloš Mulač — Tue, 26 Feb 2008 19:04:39 +0100

cluster hydra will be switched off on Wednesday from 20:00 to 21:00 due to power outage.

Cluster nympha in drain mode

Miloš Mulač — Mon, 25 Feb 2008 09:50:24 +0100

Cluster nympha temporarily does not accept new jobs due to frontend failure.

electroinstallation repairs at RUK

Michal Voců — Fri, 22 Feb 2008 13:36:14 +0100

servers acharon, dali and AFS servers for cell ruk.cuni.cz will be unaccessible due to the planned reconstruction of electroinstallation from Friday 23rd, 6pm till Saturday 24th, 6pm.

Breakdown of nympha frontend

Miloš Mulač — Mon, 18 Feb 2008 12:36:40 +0100

frontend of cluster nympha was down during weekend which caused that home directories on the cluster were unreachable. It may negatively influence running and starting jobs. Please, resubmit problematic jobs if necessary. We apologize for inconvenience.

acharon upgrade

Michal Voců — Thu, 14 Feb 2008 12:35:41 +0100

server acharon.ruk.cuni.cz will be taken offline on Friday at 9am due to planned upgrade of the OS. We expect to have it back online in the afternoon.

server mat is down

Michal Voců — Tue, 12 Feb 2008 09:48:03 +0100

server mat.ruk.cuni.cz is unaccessible due to the power outage and subsequent UPS failure. We expect to have it back online on Wed 13th, 2008.

Cluster perian fully available again

Zdenek Salvet — Thu, 7 Feb 2008 11:12:23 +0100

Cluster perian has been put to production again. Its 32-bit nodes (except those with old running jobs) have been upgraded to new version of operating system - Debian 4.0. Such nodes can be selected in PBS using "debian40" property. In spite of our care to achieve backward compatibility some differences can influence your applications. Please report any problems as soon as possible.

Cluster perian out of service - new operating system installation

Zdenek Salvet — Thu, 31 Jan 2008 07:13:58 +0100

nodes of cluster perian will be out of service today due to installation of new version of operating system (Debian 4.0). We will not kill current running jobs, the reinstallation of busy machines will be performed upon completition of those jobs.

New disk space for home directories on VOCE UI

Tomas Kouba — Fri, 18 Jan 2008 14:30:39 +0100

a new disk space has been added to skurut4 (glite User Interface) and dedicated for user's home directories. /home is now a NFS volume with 4.8T space. User's data has been migrated to the volume. Old home directories are still accessible in /home_old. We kindly ask users to delete their unwanted files in /home_old. In case of problems please contact https://rt3.cesnet.cz/rt/ On behalf of VOCE -- Tomas Kouba

network outage in Plzen

Martin Kuba — Thu, 10 Jan 2008 11:04:40 +0100

a rearrangement of the CESNET network node in Prague will start on Friday 11th January at 22:00. It will mainly affect Plzen, where two 15-minutes outages are expected. Other lines should be operational. The end is planned at 6:00 on Saturday morning.

filesystems

Martin Kuba — Thu, 20 Dec 2007 15:48:19 +0100

The portal entry 'My account - Change quota' was replaced by a new entry 'Filesystems', which displays how networked filesystems are shared among machines. For example you can find that your home directory on machines of the cluster skirit is shared with machines of clusters manwe and perian, but not with other clusters. Or that all machines in METACentrum share your AFS home directory. You can ask for the change of a filesystem quota in the same way as before.

Jobs deletion

Miloš Mulač — Sat, 17 Nov 2007 11:34:52 +0100

because of lost of connectivity in Cesnet network, some jobs were deleted on clusters konos and skurut. We appologize for problems arised.

Planned outage of file services on cluster skurut

Martin Kuba — Tue, 6 Nov 2007 17:40:27 +0100

there will be a planned outage on 14. November 2007 morning, lasting 5 to 10 minutes, of the machine providing file services to the cluster skurut. Running jobs may be affected because of this, despite that we will try to do our best so that they will not be affected.

forgotten password change

Martin Kuba — Wed, 24 Oct 2007 13:27:01 +0200

it happens sometimes, that a METACentrum user forgets his or her password. The change of password is a tricky thing, because a potential attacker may abuse it for getting access to user's data. That's why it was necessary up to now for such a user to come personally to one of our contact places. Now we want to make it more easy, so there is a new procedure. If you are from Charles University, University of West Bohemia, Masaryk University, wi-fi federation Eduroam, or from an institution involved in the Czech Identity Federation (CESNET, FEL ČVUT, MU, UK, ZČU, MNUL), or you have a trustworthy digital certificate, you can register you identity in that institutions in the portal section My Account/Personal information. When you forget your password, you can log into METACentrum portal using the registered identity and change you password in the section My Account/Password change. Of course, it is necessary to make the registration *before* you forget your password. If you cannot register any of that identities, then please keep your phone number up-to-date. When you forget your password, you can create a request for a password change in the portal section My Account / Change password. We will confirm by calling that phone number that the request was made by you.

Power outage

Miloš Mulač — Thu, 27 Sep 2007 10:56:47 +0200

Due to sudden power outage some clusters (konos, minos, nympha and hydra) and services were unavailable. If you run some jobs on there clusters, please check them and resubmit if necessary.

Cluster skurut - short service cut-off

Miloš Mulač — Wed, 26 Sep 2007 13:02:08 +0200

Cluster skurut will be switched off on Thursday 4.10. morning due to works on wiring. Now, it is not possible to submit jobs through queue long, the day before outage all submits will be blocked. The cluster operation will be switch back to normal after the wiring revision.

Planned outage of cluster skurut

Martin Kuba — Wed, 26 Sep 2007 12:53:55 +0200

the skurut cluster will be down on October 4, 2007, due to planned changes on power cables. The outage will probably last 4 hours. During that time the METACenter portal will not be available as well.

connection outages and holded jobs

Martin Kuba — Mon, 17 Sep 2007 13:09:25 +0200

there were connection outages during the last weekend at the ICS MU computing center. The connection problems led to problems in submitting jobs in the whole METACenter and to putting some jobs on hold on the skirit cluster. The problems are resolved now. We appologize for the inconvenience.

Cluster hydra temporal shutdown

Miloš Mulač — Fri, 31 Aug 2007 11:51:05 +0200

Cluster hydra will be offline from 15h today till monday morning because of air conditition system upgrade.

Cluster nympha operational

Miloš Mulač — Mon, 27 Aug 2007 13:33:44 +0200

cluster nympha is operational again (after frondend CPU fan exchange and air condition system stabilization)

Random outage of skurut cluster

Ivana Halabalová — Fri, 24 Aug 2007 13:12:14 +0200

today, 24.8., a random outage of skurut cluster happened, it was caused by an unexpected power supply outage. We apologize for any problems caused by the outage.

Cluster frontends minos and nympha failures

Miloš Mulač — Mon, 20 Aug 2007 15:50:26 +0200

Frontend of cluster minos is out of order because of power supply upgrade Frontend of cluster nympha is out of order because of processor fan failure We do not recommend to submit jobs to clusters minos and nympha during the repair.

Change of parameters specifying the number of CPU's per node

Ivana Halabalová — Tue, 7 Aug 2007 15:46:44 +0200

The new node parameters specifying the number of CPU's per node -- 2cpunode, 4cpunode etc. have been changed (technical reasons) to nodecpus2, nodecpus4. Thanks for understanding.

Planned outage of clusters in Plzen in 14.8-17.8.

Ivana Halabalová — Tue, 7 Aug 2007 14:18:18 +0200

There will be a maintenance outage of clusters located in Plzen in 14.8-17.8. (may be prolonged till 19.8.). Clusters, kerberos servers, licence and AFS servers will not be available due to large changes in the network. We do not recommend using clusters in Plzen (konos, minos, hydra, nympha, ajax) in these days. Thanks for understanding and we apologize for possible difficulties.

unerachable home directories

Ladislav Havlík — Mon, 6 Aug 2007 11:27:03 +0200

the disk failure at the machine skirit.ics.muni.cz caused (except the previously announced AFS failure) also unreachable home directories shared on the cluster skirit.

AFS failure

Ladislav Havlík — Mon, 6 Aug 2007 11:07:55 +0200

there was a hw problem on Saturday (4.8.2007) at 5:00(AM) with the disk on the AFS server. That caused problem with home directories (namely "/home/login/shared"). Some PBS jobs may have been canceled because of it. The problem was resolved on Sunday morning. We apologize for any inconvenience, please resubmit canceled jobs.

New node properties in PBS

Ivana Halabalová — Fri, 3 Aug 2007 14:00:33 +0200

New node properties have been added to PBS. They specify the number of CPU's per node. For example 2cpunode stands for 2 CPU's per node, 4cpunode stands for 4 CPU's per node etc.

qfree command behaviour change

Miloš Mulač — Mon, 30 Jul 2007 11:28:19 +0200

- the command was modified to return the same results as monitoring web page http://meta.cesnet.cz/pbsmon/nodes.do - now, the commnad syntax is: qfree [-v|-vv] qfree -l node_property[:node_property...] [-q queue_name] [-v] qfree -i Returns list of free PBS resources. -v, -vv increase verbosity level -l node_property print free nodes with properties PROP properties are separated by colon -q queue_name name of queue via which user wants access nodes -i print free licences - the command output was extended to return list of queues which may be used for job sumission; it implies a bit different format of output, which may be important if one uses the command in own scripts - the output is personalized, it means it may return different outputs for different users depending on what nodes they can really access - now user can specify the queue name in query (-l) for free nodes (i.e. if he/she wants to know how many free nodes to which it is possible to submit throught queue normal are in Prague, he/she may invoke 'qfree -l praha -q normal'; if he/she needs to know also node names, option '-v' may be added)

Clusters minos, konos and nympha operational

Miloš Mulač — Fri, 27 Jul 2007 11:31:12 +0200

Clusters minos, konos and nympha are fully operational again.

Forced shutdown of pilsen clusters

Miloš Mulač — Wed, 25 Jul 2007 11:05:26 +0200

cluster minos, konos and nympha had to be shutted down due to air conditioning and power failure.

METACenter user database maintenance

Aleš Křenek — Wed, 25 Jul 2007 09:25:42 +0200

Installation of security patches on the METACenter users database server is planned for morning, July 26. Related web applications (request for accounts, registration processing etc.) will not be operational for approx. 2 hours. The maintenance does not affect operation of computational nodes and storage access. -- Ales Krenek

RT system

Martin Kuba — Thu, 19 Jul 2007 17:29:26 +0200

The METACentrum uses a user request tracking system. All new e-mails received at the address meta@cesnet.cz are recorded in our RT (Request Tracking) system, which creates so called ticket and assigns a number to it. Creation of a new ticket is announced to the author of the original e-mail and also to all METACentrum people responsible for dealing with user requests. The ticket number is in the subject field of all e-mails concerning the ticket, for example [CESNET #12345]. The ticket number is used for matching e-mails received at meta@cesnet.cz with their corresponding tickets. So please keep this number in e-mail subjects. When a ticket is solved, it gets closed. The original author is notified about it by e-mail. Any subsequent reply again opens the closed ticket. The RT system ensures that no e-mail received at the meta@cesnet.cz address gets unnoticed.

acharon.ruk.cuni.cz is down

Michal Voců — Sun, 15 Jul 2007 16:50:51 +0200

- we were forced to turn off the server acharon.ruk.cuni.cz due to the cooling problems.

Failure in cell AFS ics.muni.cz and skirit:/home

Zdenek Salvet — Mon, 9 Jul 2007 14:45:23 +0200

AFS volumes and /home located on skirit.ics.muni.cz were unavailable due to disk array failure. They should work correctly again now.

Cluster minos operational again

Miloš Mulač — Wed, 4 Jul 2007 10:20:52 +0200

Cluster minos is now operational again. OS was upgraded to Debian Sarge. Frontend architecture is AMD Athlon, the same as architecture of workernodes now. In the case of any problems, please, let us know to meta@cesnet.cz

Cluster nympha operational again

Miloš Mulač — Thu, 21 Jun 2007 13:11:31 +0200

Cluster nympha is now operational again. OS was upgraded to Debian Sarge. In case of any problems, please, let us know to meta@cesent.cz

Nympha cluster frontend reinstallation

Miloš Mulač — Fri, 15 Jun 2007 14:07:05 +0200

today, OS of frontend of cluster nympha is being reinstalled. We assume it will be opperational again later in the afternoon.

server mat.ruk.cuni.cz will be offline

Michal Voců — Fri, 15 Jun 2007 10:01:09 +0200

Server mat.ruk.cuni.cz will be taken offline due to the power grid reconstruction on Friday, June 15th at 19:00. We expect the server to be back online on Monday, June 18th.

Personal view of PBS

Martin Kuba — Thu, 3 May 2007 10:30:54 +0200

A new menu item "Personal view" was added to the METAPortal menu section "Current state - PBS status". It displayes queues and computing nodes accessible to a user. Access to the page is controlled using the same mechanism as in the section My Account, so you have the same choice of password, certificate or Kerberos ticket.

Failure of home folders at skirit

Jan Kmunicek — Wed, 18 Apr 2007 10:30:33 +0200

Tonight there was a failure of disc array with home folders and software at ICS MU Brno similar to the failure one week ago. The access to home folders in /home and AFS volumes has been already recovered. Please, check state of your files that have been changed by you during last days especially in AFS volumes at skirit (location of folder in AFS can be obtained by command fs whereis). We can not guarantee that there will be no further problems before final solution of the failure, therefore we recommend you to minimally use directly /home in your jobs. Moreover, formulate your jobs in such a way that they will work only in /scratch after copying the input data (this is generally recommended approach if the amount of input data is not too large).

Failure of disc field at ICS MU

Jan Kmunicek — Tue, 10 Apr 2007 17:55:08 +0200

There was an unexpected failure of disc field containing home folders and software at ICS MU. Currently, the problem has been already repaired. We are terribly sorry for any inconveniences caused!

Expansion of METACentrum hardware portfolio

Jan Kmunicek — Fri, 16 Mar 2007 13:23:22 +0100

METACentrum hardware portfolio has been currently expanded by new machines that are part of cluster skirit - nodes skirit49 to skirit84. Each node is formed from 2 dual-core CPUs with 3.5 GB memory (CPU Intel Xeon 5160 "Woodcrest" 64 bit with 4 MB cache shared by two cores). Machines skirit49-1 to skirit84-1 are already available for standard utilization by all users. Corresponding mirror images skirit49-2 up skirit84-2 are virtual machines running at the same physical machines and currently serve for administrators testing purposes. New nodes of skirit cluster are accessible in PBS through property "em64t" that is principally compatible analogy of property "amd64". Simultaneously with installation of new nodes there were created new properties of computational nodes. We create a new property "x86_64" that corresponds to 64 bit PC-compatible architecture (it is unification of machines with properties "amd64" and "em64t"). Moreover, we created property "x86" that corresponds to all machines corresponding to 32 bit PC-compatible architecture (it is unification of machines with properties "x86_64" and "i386"). Although we spared our effort in customizing the environment of newly available computational capacities to be fully functional, it is possible that you encounter potential errors. In such a case we kindly ask you for your tolerance and report of found problems to address meta@cesnet.cz With regards and wishes of pleasant META computing Jan Kmunicek

Cluster minos and nympha shutdown

Miloš Mulač — Mon, 12 Mar 2007 17:02:34 +0100

Cluster minos and nympha were turned off because of air-conditioning breakdown. Nodes with running jobs as well as front-end nodes will stay running.

Cluster skurut operational again

Jan Kmunicek — Mon, 12 Mar 2007 13:20:26 +0100

Cluster skurut is operational again. OS was upgraded to Debian 3.1 Sarge. Please, report any problems you encounter to meta@cesnet.cz.

OS upgrade on cluster skurut

Miloš Mulač — Fri, 23 Feb 2007 13:19:25 +0100

We would like to start reinstalations of cluster skurut in the second half of the next week in order to upgrade of OS to Debian Sarge. All data on computational nodes (/scratch/*) will be lost. On the other hand, data on volume skurut:/home will persist. We ask all owners of running jobs to stop them. Users will be informed via e-mail when cluster is operaional again.

Connection outage at UVT MU

Martin Kuba — Tue, 20 Feb 2007 16:05:46 +0100

There was an outage of the Internet connection at UVT MU today between 1 and 2 PM. This can cause a failure of some jobs in PBS.

Short-term power outage of aule machine

Jan Kmunicek — Wed, 31 Jan 2007 13:42:57 +0100

There was a short-term power outage of aule machine. It is not yet clear if the outage was caused by software error or hardware problem. Currently, the machine is again up and running.

Extension of METACentrum account

Jan Kmunicek — Mon, 4 Dec 2006 11:13:54 +0000

Concerning upcoming end of year 2006, do not forget to renew your METACentrum account. The way is the same as it was during last years, e.g. you have to visit pages http://meta.cesnet.cz, authenticate yourself at My account part of the portal and you will add your report of your performed work during year 2006 in section Annual report and extension of accounts. Your report will be used as a base for renewal of your account in METACentrum. If you do not submit your report until the 31. 12. 2006 your METACentrum account will be blocked from 1. 1. 2007. Please, check also that your registered personal information are still valid. If you have an electronic certificate issued by CESNET CA or you have at your home institute activated service Eduroam, we suggest you to register corresponding information to your personal data. Using them you will obtain an access to some METACentrum services (e.g. change of forgotten password).

EGEE II project seminar

Jan Kmunicek — Wed, 22 Nov 2006 11:55:58 +0100

Let me invite you to participate in "EGEE II project seminar", that is organized by CESNET institute and will be held on December 12. The seminar will be focused on demonstration how to use grid environment of European EGEE II project and national grid project METACentrum respectively for solving of computationally/data intensive problems together with presentation of advantages of these environments. Detailed information including registration can be found at following webpages: http://egee.cesnet.cz/en/events/fourth.html On behalf of the organiyers, Jan Kmuníček

New nodes in cluster konos

Miloš Mulač — Mon, 20 Nov 2006 12:40:05 +0100

New worker nodes konos1-10 were added to cluster konos. See detailed HW description at http://meta.cesnet.cz/cms/opencms/cs/docs/hardware.html#konos1-10

Failure of cluster nympha frontend

Miloš Mulač — Mon, 20 Nov 2006 12:19:58 +0100

nympha.zcu.cz was down because of OS failure during this weekend. It caused temporal inaccessibility of home directories on the cluster nympha.

afs cell ruk.cuni.cz is down

Michal Voců — Thu, 16 Nov 2006 16:57:04 +0100

The AFS cell ruk.cuni.cz is down due to the disk array failure. We are working on bringing it back online.

PBS server failure

Miloš Mulač — Mon, 13 Nov 2006 17:48:59 +0100

Some jobs were accidently resubmitted during upgrade of PBS this aternoon. The jobs should be restarted automatically but we recommend to check your jobs for any anomalies. We appologize to all involved. Best regards, Milos Mulac.

Shared /home folder at aule machine

Jan Kmunicek — Tue, 7 Nov 2006 09:47:03 +0100

Machine aule was transformed into the same mode as the rest of amd64 machines, i.e. aule has currently shared home folder from machine skirit similarly as cluster manwe. Shared volume /home was created at aule machine. Due to this change a volume /scratch was enlarged as it was up to now shared with /home. Original old folder /home is temporarily available in folder /home.old. Currently the machine aule is again in PBS accessible for users.

New cluster manwe

Jan Kmunicek — Fri, 27 Oct 2006 13:07:24 +0200

Hardware portfolio of METACentrum has been extended by cluster manwe - three 16-CPUs machines Sun Fire X4600. Detailed hardware configuration can be found at http://meta.cesnet.cz/cms/opencms/en/docs/hardware.html#manwe address. Nodes manwe1 and manwe2 are freely available to all users in the same mode as machine aule, i.e. through queue normal for 1-CPU jobs and through queue parallel for parallel jobs. Unlike aule home folder is shared from machine skirit. Machine manwe3 is reserved for internal tests of METACentrum administrators. Best regards and happy META computing, Jan Kmuníček

Reinstallation of cluster puu

Miloš Mulač — Tue, 26 Sep 2006 09:31:52 +0200

It is not possible to start new jobs on cluster puu since yesterday's evening. The cluster will be reinstalled after running jobs finish and users will be informed when it is ready.

Restoration of AFS finished!

Jan Kmunicek — Mon, 18 Sep 2006 15:05:50 +0200

The announced restoration of AFS filesystem in METACentrum has been successfully completed. We would like to still kindly ask you to report any potential persisting problems at meta@cesnet.cz address.

Cluster Konos start-up

Miloš Mulač — Mon, 18 Sep 2006 13:11:58 +0200

we have started new cluster Konos. The name of front-end is konos.fav.zcu.cz. Cluster consist of approx. 100 AMD opteron processor cores arranged into dual singlecore or dual dualcore nodes, each with 4GB of memory. (detailed information can be found at METACentrum web pages, in section Hardware - https://meta.cesnet.cz/cms/opencms/en/docs/hardware.html#konos-fr) The cluster is owned by Department of Mathematics and Department of Cybernetics of University of West Bohemia (group ITI). Common users can access the cluster via PBS queues 'short' or 'normal'. Members of ITI may use queue iti. In the case you meat any problems dealing with the cluster, please, let us know at address meta@cesnet.cz.

AFS problems

Jan Kmunicek — Wed, 13 Sep 2006 09:54:42 +0200

There was a failure of AFS filesystem in the whole METACentrum. Therefore the AFS home folders are not available and you can encounter problems with non-functionality of other services too. We would like to kindly ask you to report any problems related to non-availability of software packages or user data at meta@cesnet.cz address. We are terribly sorry for any inconveniences caused, the problem is solved with maximal priority.

Power outage in Pilsen

Miloš Mulač — Tue, 12 Sep 2006 16:39:01 +0200

we have had to switch off clusters minos, nympha, puu and konos because of power outage. We will try to start the clusters during afternoon.

Cluster skirit fully functional again!

Jan Kmunicek — Fri, 25 Aug 2006 17:17:02 +0200

Cluster skirit has been completely reinstalled and is currently fully functional again and running new version of operation system. Moreover, new version of key applications has been installed - Maple-10.04 and patched Amber-8.0 (patch level 63, parallel versions should be run using mpiexec). We also unified different versions of Myrinet network interface into unified Myrinet2000 (currently 28 nodes). We would like to kindly ask you to report any potential problems or drawbacks to allow us to solve them quickly and efficently!

Unexpected outage of cluster skurut

Jan Kmunicek — Mon, 21 Aug 2006 12:32:49 +0200

There was an unexpected outage of skurut cluster at CESNET computer hall caused probably by power failure. We are sorry for any inconvenience caused!

Reinstallation of RB at skurut3; skurut4 disk space extension

Jiri Kosina — Tue, 8 Aug 2006 11:48:58 +0200

Tomorrow (Wednesday 09 Aug 2006) afternoon, the VOCE Resource Broker on skurut3.cesnet.cz will be reinstalled, and disk space will be extended in skurut4.cesnet.cz (VOCE User Interface). Expected downtime is 13:00 - 18:00. For submitting jobs, skurut2.cesnet.cz (gLite WMS for VOCE) still can be used. Thanks for understanding and sorry for any inconveniences caused.

Modifications of computational resources at Brno's node

Jan Kmunicek — Thu, 27 Jul 2006 11:21:27 +0200

There is available cluster perian together with frontend of skirit cluster (its disc capacity of home directory was substantially enlarged) after planned power outage at Brno's METACentrum node. Individual worker nodes of cluster skirit are not available yet due to operation system upgrade. If you urgently need to submit computational jobs, use computational capacities available in Prague and Pilsen. Thank you for your understanding!

Power failure of METACentrum Pilsen node

Jan Kmunicek — Mon, 24 Jul 2006 07:38:42 +0200

There was an unexpected power failure of all clusters and machines at Pilsen METACentrum node on July 21 as a result of air-conditioning breakdown. Expected time of operation restoration is tomorrow afternoon.

acharon.ruk.cuni.cz turned off

Michal Voců — Thu, 20 Jul 2006 16:21:28 +0200

Due to the air-condiotioning failures we were forced to turn off the server acharon.

Planned power outage of SCB MU

Jan Kmunicek — Thu, 13 Jul 2006 10:58:46 +0200

On July 26 there will be an installation of new UPS system followed by modification of power circuit at computer hall at SCB MU in Brno carried on. Due to this work the whole METACentrum node in Brno will not be functional on July 26 - all UPS will not be running including backup diesel assembly. We apologize in advance for any inconvenience caused!

acharon.ruk.cuni.cz reboot

Michal Voců — Thu, 29 Jun 2006 12:32:40 +0200

The server acharon.ruk.cuni.cz stopped resonding, we were forced to reboot it.

acharon upgrade

Michal Voců — Mon, 19 Jun 2006 13:44:54 +0200

server acharon.ruk.cuni.cz will be taken offline for hardware upgrade on June, 21st.

reboot of nympha.zcu.cz

Miloš Mulač — Fri, 9 Jun 2006 13:28:08 +0200

memory page allocation arror; It was necessary to reboot the computer. The failure could influence running jobs, especially if they were using shared volume nympha:/home.

Summer school "Structure and function of biomolecules"

Jan Kmunicek — Mon, 22 May 2006 05:09:31 +0200

This news does not concern non-Czech speaking users as the announced school will be in Czech language only.

Expansion of METACentrum hardware portfolio

Jan Kmunicek — Fri, 12 May 2006 00:43:11 +0200

METACentrum hardware portfolio has been currently expanded by new machines. Machine acharon.ruk.cuni.cz that represents SGI Itanium 2 architecture and machine aule.ics.muni.cz built on dual-core AMD Opteron processors. Besides, there is newly available also cluster puu.fav.zcu.cz that was funded by research centre ITI ( KMA ZCU respectively). It is available to all METACentrum users through the queues 'short' and 'normal'. There is a possibility to obtain access to queue 'iti' allowing long running jobs for users participating in the project ITI. To obtain access to this queue you have to send a request to address rkuzel@kma.zcu.cz. Frontend of puu cluster (puu.fav.zcu.cz) has currently i386 architecture while the rest of the cluster is AMD64. Similarly it is possible to utilize new supercomputer ajax.zcu.cz that is available for all METACentrum users through the queues 'short' and 'normal'. There is a possibility to obtain access to the queue 'zsc' allowing long running jobs for ZCU employees and students. To obtain access to this queue you have to send a request at address knourek@civ.zcu.cz. Visit hardware section of METACentrum portal (http://meta.cesnet.cz/en/docs/hardware.html) for detailed information concerning newly available computational resources. The list of available application programs is accessible using command 'module av' at mentioned machines. Although we spared our effort in customizing the environment of newly available computational capacities to be fully functional, it is possible that you encounter potential errors. In such a case we kindly ask you for your tolerance and report of found problems to address meta@cesnet.cz With regards and wishes of pleasant META computing Jan Kmunicek

Power switch-off at May 12, 12:00 -- 17:00

Jan Kmunicek — Thu, 11 May 2006 15:56:36 +0200

Due to the power switch-off at May 12, 2006 between 12:00 and 17:00, supercomputer ajax and PC clusters puu, minos and nympha will not operate. We apologize for any problems. Jindrich Knourek, Westbohemian supercomputing center

Upgrade of skirit frontend

Jan Kmunicek — Wed, 26 Apr 2006 16:11:49 +0200

Non-planned complete power failure of Brno's node was used for operation system upgrade of skirit frontend. Other nodes are dependent of its shared file system and later upgrade would require their switch off. Currently there is a little different environment at node skirit.ics.muni.cz compared to common nodes (those were not affected). It is therefore possible that some applications will not be fully functional until debugging of all errors at node skirit.ics.muni.cz. We apologize for any inconveniences caused.

Power outage at ICS MU

Zdenek Salvet — Fri, 14 Apr 2006 17:14:01 +0200

Problems during works on upgrade of central UPS system of computer hall at ICS MU forced us to take most computing resourced located there offline till the end of UPS upgrade.

Potential power failure at ICS MU

Jan Kmunicek — Thu, 13 Apr 2006 10:16:45 +0200

There will be on-going work on reinforcement of central backuped power for computer hall at ICS MU on April 14, 2006 from 3 p.m. up to 4 p.m. Although there should be no power failure, we prefer to warn you in advance.

acharon.ruk.cuni.cz is back

Michal Voců — Fri, 31 Mar 2006 12:42:26 +0200

the server acharon.ruk.cuni.cz is operational after upgrade.

upgrade of acharon.ruk.cuni.cz

Michal Voců — Tue, 28 Mar 2006 13:59:46 +0200

the server acharon.ruk.cuni.cz will be upgraded to new OS version on Wednesday, 29th. The server will be unavailable for the time required to upgrade OS and dependent applications. We apologize for the inconvenience.

air-conditioning failure at RUK

Michal Voců — Sat, 25 Mar 2006 19:13:40 +0100

we were forced to shut down servers acharon.ruk.cuni.cz and dali.ruk.cuni.cz due to the excessive temperature caused by air-conditioning failure. We expect both servers to be back on Monday, we apologize for the inconvenience.

Power shortage

Miloš Mulač — Wed, 8 Mar 2006 11:07:04 +0100

Because of power distributor upgrade, cluster minos and nympha will be temporarily switched off.

Unplanned power failure at SCB MU

Jan Kmunicek — Wed, 15 Feb 2006 09:31:09 +0100

There was an unexpected power failure at SCB MU computer hall. Please, accept our apologies.

expected failure of /scratch

Michal Voců — Fri, 10 Feb 2006 11:35:40 +0100

The /scratch filesystem on server mat is suspected to fail. Please, copy your important data till the 17th of February; we will replace the failing disk and recreate the filesystem on that day, unless hardware failure occurs sooner.

Switching of power distribution at CESNET

Jan Kmunicek — Mon, 23 Jan 2006 08:50:10 +0100

On January 28 2006 there will be a planned outage of cluster skurut in the morning in duration of several hours due to switching of power distribution at CESNET.

upgrade of server acharon

Michal Voců — Tue, 20 Dec 2005 17:33:02 +0100

server 'acharon' was upgraded to 12 Intel Itanium 2 CPUs and 48GB of RAM.

Extension of accounts and METACentrum Yearbook

Jan Kmunicek — Thu, 15 Dec 2005 23:13:07 +0100

Dear METACentrum users, concerning upcoming end of year 2005, do not forget to renew your METACentrum account. The way is the same as it was during last years, e.g. you have to visit pages http://meta.cesnet.cz, authenticate yourself at My account part of the portal and you will add your report of your performed work during year 2005 in section Annual report and extension of accounts. Your report will be used as a base for renewal of your account in METACentrum. If you do not submit your report until the 31. 12. 2005 your METACentrum account will be blocked after 1. 1. 2006. METACentrum simultaneously plans to release METACentrum Yearbook as soon as possible (i.e. still in winter time) whose major part will be formed from your reports as usual. On one hand we need your concise report to be used for extension of your account and on the other hand we need also detailed text describing your research and results that you have achieved using computational or other METACentrum resources (this also covers situation when you use services of only one METACentrum node in a long-term or exclusive manner). This detailed text will also make tables, graphs and figures available. We expect that the Yearbook will contain only information from year 2005. Nevertheless, if you have available significant results from year 2002--2004, you can include them into your report too but principally the Yearbook will map year 2005. Basic instructions concerning formatting of your contribution for the Yearbook are available at http://meta.cesnet.cz/en/state/actual/index.html to make the text preparation easier for you (and to minimize our work with re-typesetting). We recommend you to send us the report for the Yearbook together with short report about METACentrum utilization (report for the Yearbook replaces concise report, of course), nevertheless the collection of the Yearbook contributions will stay opened after New Year's Day as well. Please, remember that when we have to consider priorities or special requests we are trying to take into account quality and extent of results you have achieved during last time period with help of METACentrum resources. The contribution to METACentrum Yearbook together with the list of your publications and presentations achieved using METACentrum is a natural way how to inform also us about the range and the meaning of your work. With best regards, Jan Kmunicek

Power outage at SCB

Jan Kmunicek — Wed, 7 Dec 2005 10:55:33 +0100

There was a complete non-planned power outage at Supercomputing Centre Brno on December 7.

Seminar for users of EGEE Grid

Jan Kmunicek — Sun, 4 Dec 2005 16:14:17 +0100

Dear METACentrum users, let me invite you to planned (the third one already) EGEE (Enabling Grids for E-sciencE) project seminar organized by CESNET association. This seminar freely follows two previous EGEE seminars. The primary aim of this seminar is to inform broad user community of high-performance/throughput computing with actual state concerning solving of the EGEE project and with planned follow-up project EGEE II, respectively. You will have the possibility to obtain information what services (the EGEE Grid infrastructure itself as well as different ways how to access it including tools for work in the EGEE Grid environment) are currently offered and available in the frame of the EGEE project. The seminar will be held on December 13, 2005 in classroom no. 304 (3th floor, block A4) of Research and Development Centre, Faculty of Electrical Enginnering, Czech Technical University from 9 a.m. with , expected end approximately at 6 p.m. The program of the seminar is planned as follows: 08.30-09.00 Registration of participants 09.00-09.30 Introduction to Grids utilization 09.30-10.00 VOCE Status 10.00-10.15 Break 10.15-12.00 CHARON System (theory & practical hands-on) 12.00-13.00 Lunch 13.00-13.30 Distribution of HW tokens and CA CESNET certificates 13.30-15.00 P-GRADE portal (theorye & practical hands-on) 15.00-15.30 Break 15.30-17.00 P-GRADE portal (theory & practical hands-on) 17.00-17.30 Logging and Bookkeeping service Therefore we would like to kindly ask you for registration at http://egee.cesnet.cz/en/events/third.html, because the number of participants is needed for arrangement of the catering. To be able to modify the available grid environment according to your - the end user - needs we we would like to ask all potential users of the EGEE Grid capacities to answer several simple questions that will help us to better understand what you expect from grid environment. 1. Which application areas are crucial for you? 2. Which application or other programs do you commonly use? 3. Do you prefer work using command line or utilization of graphical user interface, possibly intermediation through a portal? We expect your answers together with your other suggestions and requests at egee@cesnet.cz. I am looking forward to meet you on Tuesday, December 13 Best regards, Jan Kmunicek

Pushing grond and eru of the running

Jan Kmunicek — Fri, 11 Nov 2005 10:36:13 +0100

Dear METACentrum user, let us inform you that computers grond and eru have reached the end of their functionality period and they are ready to be pushed of the running. Concerning last power outages in supercomputing centre in Brno both machines are in minimal operation mode and they are removed from PBS. Therefore it is not possible to submit new jobs to them. Content of home and scratch0 folders at both machines were moved and is available at /afs/ics.muni.cz/odstavene where are prepared folders eru/home eru/scratch0 grond/home grond/scratch0 that contain copied data with proper setting of individual access rights, simultaneously we are ready to help you with transition to other METACentrum computational resources (e.g. machine mat in Prague if you need the architecture with MIPS processors). However to succeed in this aim we need information which application programs have you used at grond and eru to provide their migration to locations where those programs are not available yet. Please, send all requests regarding the pushing grond and eru of the running to RT system using the address meta@cesnet.cz With apology for potential problems that will be caused by this action and simultaneously with wish of further successful work in the environment of METACentrum Jan Kmunicek

Switching of diesel assembly 12-13. 11.

Jan Kmunicek — Tue, 8 Nov 2005 15:32:04 +0100

Dear users, we would like to inform you that no computational resources in Brno will be available on November 12-13 as a result of planned switching of diesel assembly. Therefore cluster skirit and perian will not be available. We recommend you to finish your long jobs (if possible), reschedule them to some other resources or simply do not submit them during this period. AFS and PBS servers will be fully functional.

Power outtage

Miloš Mulač — Wed, 26 Oct 2005 17:55:41 +0200

Because of power outtage combined with UPC system disfunction, cluster skirit, perian and quark, as well as some AFS volumes in cell ICS.MUNI.CZ were inaccessible. We are still working on problem correction.

minos - upgrade

Miloš Mulač — Thu, 20 Oct 2005 15:20:12 +0200

Amount of memory was upgraded from 0.5GB to 2GB and amount of disk space for partition /home was upgraded from 32GB to 340GB in server minos.zcu.cz. Cluster is operational again.

HDD upgrade

Miloš Mulač — Thu, 20 Oct 2005 10:38:11 +0200

Cluster minos will be temporarily shut down because of hdd upgrade.

PBS server problems

Miloš Mulač — Tue, 11 Oct 2005 20:30:14 +0200

Because of failure of one HDD was inaccessible PBS server during last night. It caused breakdown of several PBS jobs.

Cluster minos up again

Jindřich Kňourek — Fri, 30 Sep 2005 13:03:17 +0200

Dear users, you can use cluster minos again. For the moment, GigaEthernet switch was replaced with a 100Mbit one till the original switch will be repaired. For this reason, the power of some applications, data transfers etc can be lowered. Jindrich Knourek, WestBohemian supercomputer center.

Network down for cluster minos

Jindřich Kňourek — Thu, 29 Sep 2005 15:40:20 +0200

Dear users, due to the network switch failure it is not possible to use PC cluster "minos". We are sorry for any problems. Jindrich Knourek, Westbohemian supercomputing center

HW upgrade of frontend of cluster minos

Miloš Mulač — Mon, 5 Sep 2005 11:28:59 +0200

Dear users, because of HDD upgrade, cluster minos is temporarily off-line. We suppose it will be operational again on 6.9. afternoon. Regards, Milos Mulac.

Switch-off of clusters nympha and minos in 24.8.-25.8.

Miloš Mulač — Tue, 16 Aug 2005 13:36:54 +0200

Dear users, clusters minos and nympha will be switched off in afternoon of 24th because of recontruction of transformer station. They should be switched on next day (25.8.) in the morning. Sincerely, Milos Mulac.

METACentrum portal modification

Jan Kmunicek — Wed, 13 Jul 2005 14:11:51 +0200

We would to announce that a quick guide-post was added to the introduction page of METACentrum portal. This will allow much easier navigation to selected key information (current state of resources, personal settings etc.).

Cell ruk.cuni.cz down

Zdenek Salvet — Mon, 11 Jul 2005 11:17:08 +0200

Kerberos and AFS server in the Supercomputer centre of Charles University has failed. Automatic autentization of running jobs to AFS cell ruk.cuni.cz has been switched off. Duration of hte failure is not know yet.

SC UK is operational

Michal Voců — Sun, 3 Jul 2005 09:52:42 +0200

Servers at SC UK are operation after power outage causing failure of several hard disks. The one exception is hal.ruk.cuni.cz with failed system disk which must be reinstalled.

Electricity failure.

Jan Kmunicek — Thu, 30 Jun 2005 13:28:17 +0200

Due to problems with UPS there was a electricity failure at perian cluster (concretely machines perian61-68, perian70, perian71).

News and outages as RSS

Martin Kuba — Thu, 23 Jun 2005 13:55:33 +0200

Announcements about news and outages are now available also as RSS (Really Simple Syndication) feed, i.e. special actualized XML file. RSS is available in both Czech and English. Link to RSS file is under the small "RSS" icon just below the menu. RSS can be read either in FireFox browser using its Live Bookmarks feature, or in a specialized software like FeedReader, or using web-based services (http://rss.plech.net/).

News announcements

Martin Kuba — Thu, 23 Jun 2005 13:44:34 +0200

The METACentrum portal has a new feature which allows administrators to send emails to all users, which have news announcements enabled in their profile. The default setting is that announcements are enabled. You can change it on the portal in section "My Account - Personal Information".

Ganglia tool