EMI 1 Kebnekaise Products
Full list of 56 EMI 1 Kebnekaise Products, with detailed release notes and instructions for their installation/configuration.
Products list
WMS
WMS v.3.3.0
- Release Notes
- List of RFCs
- Documentation
- Artefacts
- Updates
- WMS v. 3.3.1
- WMS v. 3.3.1-1 (Emergency release)
- WMS v. 3.3.2
- WMS v. 3.3.3
- WMS v. 3.3.4
- WMS v. 3.3.5
Release Notes
What's new
- Job replanning based on a fixed timeout. Jobs stuck at blocking queues oftentimes represent an annoying issue which has been never specifically targeted by the WMS. This new feature will allow to rematch/replan jobs after they sit in a given queue for too long. This feature is enabled setting the JDL variable EnableWmsFeedback to "true" For now, the queue removal timeout is centrally set in the WMS configuration. [WorkloadManager] ReplanGracePeriod = 1800; in the same place another parameter exists to specifiy the maximum allowed number of replans MaxReplansCount = 4; after this threshold is reached the job is aborted.
- Adopting Gridsite 1.5.19 now allows the WMS to work with RFC-style VOMS proxies. Delegation-2 protocol is also supported
- Thanks to several improvements introduced by LB 2.1, a collocation scenario between WMS and LB server is now sustainable and encouraged.
- New attribute wmsrequirements, added to the configuration file in the WM section. This expression is appended in && to the user requirements: requirements = (userrequirements) && (wmsrequirements); The default value for this attribute (set by yaim) is:
WmsRequirements = ((ShortDeadlineJob =?= TRUE) ?
RegExp(".sdj$", other.GlueCEUniqueID) :
!RegExp(".sdj$", other.GlueCEUniqueID)) &&
(other.GlueCEPolicyMaxTotalJobs == 0 ||
other.GlueCEStateTotalJobs < other.GlueCEPolicyMaxTotalJobs) &&
(EnableWmsFeedback =?= TRUE ?
RegExp("cream", other.GlueCEImplementationName, "i") : true);
- WMS now considers in the matchmaking the value published for MaxTotalJobs
- When a collection is aborted the "Abort" event is now logged for all the sub-nodes as well
- Retry policies for ISB download and OSB upload are now separated. The defaults coded in the jobwrapper template are: ISB: retry = 2 first_wait= 60 OSB: retry = 6 first_wait= 300
- All attributes of a SA/SE can now be used in gangmatching
- Starting from the present release, a new semantics, more correct and less implicit, has been introduced in the definition of fqans in the gridsite gacl file: the fqan must be valid (i.e. with the leading slash) and must exactly match the user fqan.
- Other bug fixes
Deployment notes
- Yaim (re)configuration is needed after installation/update New yaim variable GLITE_LB_TYPE Set this variable to "both" to allow the colocation between WMS and LB. Otherwise, the variable must be set to 'proxy'
Known issues
- DAGs are not supported for submission to CREAM. As typical for 64-bit architectures, the process virtual memory footprint might increase considerably. Given that, especially for WM and ICE, this was a concern even on 32-bit, an alternative malloc, as taken by the google performance tools, is being automatically deployed, as part the metapackage and configured by yaim.
- Because of bug #82083 in gridsite, VOMS server certificates are still needed (i.e. .lsc files are not enough) for EMI-1 VOMS servers (while for gLite 3.2 VOMS server this is not needed)
List of RFCs
- separate retry policies for ISB and OSB
- Not all attributes of a SA/SE coul be used in a gangmatching
- When a collection is aborted the "Abort" event should be logged for the sub-nodes as well /2
- WMS should consider MaxTotalJobs
- Optimization of resubmission
- job wrapper should log events for truncated files
- glite-wms-wmproxy to support graceful command
- WMProxy does not catch signal 25
- wmproxy.gacl person record allows anyone to pass
- yaim-wms: host{cert,key}.pem in /home/glite
- WMS 3.2 WMProxy logs are useless below level 6
- WMP finds FQAN inconsistency only if GROUPS are different, not ROLES
- WM running but not processing jobs
- the amount of information logged to the LB needs to be reviewed
- ListMatch should consider also SDJ specification
- WMProxy Server: gSoap needs to be built with WITH_IPV6 flag
- Request for a feature allowing propagation of generic parameters from JDL to LRMs
- Request for handling SMPGranularity attribute in the JDL
- limit maximum sleep time in job wrapper
- user job is not killed when proxy expires
- yaim-wms: Enable Glue 2.0 publishing
- glite_wms_wmproxy_load_monitor has a problem with lvm partitions
- the proxycache purger needs to be made compatible with the latest gridsite releases
- Bug in ICE's start/stop script
- WMS hates collections with 192 nodes!
- glite-wms-create-proxy "ambiguous redirect"
- environment values in JDL cannot have spaces
- JobWrapper tries to use "test -eq" for string comparison
- Submission failed due to a credential problem
- Wrong retry count computation
- edg_wll_SetLoggingJobProxy with empty sequence code returns "no state in DB"
- missing ReallyRunning event from LogMonitor
- Previous matches information is not taken into account if direct submission is used
- Wrong counter in ICE database is set at the job creation
- wrong reason logged
- ICE should log a DONE_FAILED to LB every time the job is going to be resubmitted
- Synchronization loss between real validity of proxy and exp. time saved in ICE's database
- Wrong myproxyserver string processing in ICE
- "MyProxyServer: wrong type caught for attribute" for parametric jobs
- Env variables, ~ character are not correctly expanded in the WMS UI
- Resource BDII for WMS needs to be revisit
- While purging DAGs/Collections the CLEAR event is only logged for the parent node
- LB Query timeout
- Alternative GLITE_WMS_LOG_DESTINATION in the jobwrapper
- yaim-wms:yaim should set IsmIiLDAPCEFilterExt according to the supported VO(s)
- YAIM_WMS:Multiple parameter configuration added in condor_config.local
- various bugs about parametric jobs
Documentation
- Functional Description
- User Guide
- Client Installation and Configuration : see EMI UI documentation
- User Troubleshooting Guide
- System Administrator Guide
- Service Reference Card
- Product factsheet
- Other Documentation