Agenda-09-02-2015

From EGIWiki
Revision as of 14:04, 9 February 2015 by Caifti (talk | contribs) (News from URT)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Audio conference link Conference system is Adobe Connect, no password required.
Audio conference details Indico page



Middleware releases and staged rollout

News from URT

Recent, or future planned, releases from the product teams:

  • APEL 1.4.0

What’s New

   Added ability to process files compressed with bzip2.
   Added multiplier retrieval to GridEngine parser for CPU and wall-clock time.
   Fixed integer overflow when loading normalised summaries into APEL servers.
   Corrected the way some logging calls were made and reduced the logging level when trying to purge a directory that isn't empty.
   Changed server authorization system to keep the list of DNs for a limited time if unable to update from GOCDB and updated GOCDB query URL.
   Added Month and Year columns to primary key of CloudSummaries table in cloud schema.

  • Globus - in EPEL Testing:

EPEL5: https://admin.fedoraproject.org/updates/FEDORA-EPEL-2015-0515/ EPEL6: https://admin.fedoraproject.org/updates/FEDORA-EPEL-2015-0524/ EPEL7: https://admin.fedoraproject.org/updates/FEDORA-EPEL-2015-0499/

UMD release

  • UMD 3.11.0 in preparation

Staged rollout updates

  • argus-ees v. 0.2.1
  • UNICORE 7.2.0 components (tsi, registry, gateway, xuudb, unicore-x)

In Verification

N/A

New Products

  • squid v. 2.7.19
  • fts3 v. 3.2.30

Ready to be released:

  • apel v. 1.3.1
  • apel-ssm v. 2.1.5
  • Blah 1.20.7
  • CGSI-gSOAP v. 1.3.7
  • CREAM v. 1.16.4
  • cream-ge v. 2.3.0
  • cream-torque v. 2.1.4
  • Davix v. 0.4.0
  • DMlite 0.7.2
  • DPM 1.8.9
  • F.SQUID v. 2.7.19-21
  • FTS3 v. 3.2.30
  • gfal2 v. 2.7.8
  • gfal2-python 1.6.0.
  • Globus 6.0.0 products: Gridftp, Globus-Default-Security, Gram5, MyProxy;
  • px v. 1.3.36
  • qcg-computing-3.4.0 (only SL6)
  • qcg-notifications-3.4.0 (only SL6)
  • srm-ifce v. 1.21.4
  • squid v. 2.7.19
  • Voms Admin v. 2.0.12
  • VOMS clients v. 3.0.5/3.0.6
  • wms-ice v. 3.5.4

UMD 3 EA

  • Some sites have the contact points for the EA adopters outdated so please check in table if all contacts and products are still correct and send me email if you need to add / remove some contacts (SSO account mandatory): (full site list)
  • If anyone interested please contact me or cristina to be included in the early adopter list.

Next releases

  • End of Jan 2015
  • 2015 Calendar is in planning phase

Operational issues

Report from DMSU

  • FYI: Collocated CREAM and ARGUS authorization issues. There were two similar issues where CREAM's queries to ARGUS were being refused. Quoting from the solution: The issue was caused by configuration of CREAM and Argus nodes with different group sets on the same machine. This is not supported by YAIM, special care is necessary. Since this problem cropped up at least twice in very short succession, it may be worth mentioning. No special action is necessary, though.

The issue is represented like this in the CREAM log (excerpt form the NecTec site):

26 Jan 2015 03:24:54,480 INFO org.glite.ce.commonj.authz.gjaf.ServiceAuthorizationChain - User CN=Liaw SyueYi 182693, OU=GRID, O=AS, C=TW not authorized for {http://www.gridsite.org/namespaces/delegation-2}getProxyReq
26 Jan 2015 03:24:54,481 INFO org.glite.ce.commonj.authz.axis2.AuthorizationHandler - request for OPERATION={http://www.gridsite.org/namespaces/delegation-2}getProxyReq; REMOTE_REQUEST_ADDRESS=117.103.109.193; USER_DN=CN=Liaw SyueYi 182693,OU=GRID,O=AS,C=TW; NOT AUTHORIZED
26 Jan 2015 03:24:54,481 ERROR org.apache.axis2.engine.AxisEngine - Authorization error
org.apache.axis2.AxisFault: Authorization error
at org.glite.ce.cream.authz.axis2.AuthorizationHandler.getAuthorizationFault(AuthorizationHandler.java:155)
at org.glite.ce.commonj.authz.axis2.AuthorizationHandler.invoke(AuthorizationHandler.java:162)
at org.apache.axis2.engine.Phase.invokeHandler(Phase.java:340)
at org.apache.axis2.engine.Phase.invoke(Phase.java:313)
at org.apache.axis2.engine.AxisEngine.invoke(AxisEngine.java:262)
at org.apache.axis2.engine.AxisEngine.receive(AxisEngine.java:168)
at org.apache.axis2.transport.http.HTTPTransportUtils.processHTTPPostRequest(HTTPTransportUtils.java:172)
at org.apache.axis2.transport.http.AxisServlet.doPost(AxisServlet.java:146)

APEL multicore accounting

As many EGI user communities are now exploiting multicore hardware it is important for them that accounting correctly reflects the usage made of cores and cpus. At the December Operations Management Board (OMB) it was decided to ask all sites using the APEL client to configure it to publish the number of cores used by jobs.

To enable multicore accounting, you will need to edit the configuration file for the apel parser. This is the software which parses blah logs and batch logs to produce accounting records. The configuration file is usually found at /etc/apel/parser.cfg. In the section labelled [batch], change:

parallel = false

to

parallel = true

This will enable multicore reporting for all future accounting data. Please note that this does not change historical data. Also note that republishing old data is not sufficient to show multicore information - the log files will need to be reparsed. If you wish to republish old data with multicore enabled, please open a GGUS ticket with the APEL team so that we can help you with the process.

If you use the SGE parser, please be aware that it only reports on the number of processors used in a job. It does not report the number of nodes. If you know how to get around this limitation, then please get in touch with the APEL team at apel-admins@stfc.ac.uk.

The multicore accounting data can currently be seen here: http://accounting-devel.egi.eu/show.php?

Drill down to your site and select the grouping "Show data for: Submitting Host" as a function of: "Number of Processors".

Values of 0 mean the parallel option was false when the data were published. The Submitting Host is a new feature in the accounting portal which lets a site see in more detail which CEs are publishing.

EMI-2 decommissioning

  • IMPORTANT - switched-off UMD2/EMI2 APEL service - Friday 16th January
  • ended

dCache 2.2.X decommissioning

  • finished


SAM Nagios probes re-factoring

  • no news

MySQL 5.0 EOL

2.8 SL/SLC/CentOS 5 Support Lifetime

AOB

Monthly Availability/Reliability

  • [1] - are followed up through Parent GGUS #111193 - still 10 tickets open:
  • ROC_Asia/Pacific, IDGF, NGI_BG, NGI_CHINA, NGI_GRNET, NGI_MARGI, NGI_IL, NGI_DE, NGI_FRANCE, NGI_RO


Next meetings

  • March 09, 2015

Minutes