My tasklist

A summary of what I did for CMS operations in the day-to-day

Workflow team support

• Read all the Workflow Team Support elogs (https://cmslogbook.cern.ch/elog/Workflow+processing/) looking out for issues that are likely WMAgent bugs or that require semi-invasive interventions in the WMAgent such as peeking/changing information in the SQL databases or harvesting information out of couchdb that is not available in the usual monitoring. Some usual issues have been components going down and not restarting properly, stuck requests, missing or not registered output data. There is one issue which will pop up frequently and it is transient errors with couchdb, note that we are very aware of couchdb instability and there are already efforts by one of the new DMWM developers, Alexander Richards, to upgrade the couchdb infrastructure. Finally, while reading all these elogs, try to catch for patterns that that indicate places where the code can be improved/optimized.
• Attend the workflow team meetings and answer questions, mostly about the functioning of the WMAgent and also the current development issues for the project. Also receive requests and feedback on the development needs from the workflow team.
• Answer support questions about the WMAgent working in the hypernews and other email groups
• Write scripts on demand that use WMCore libraries to solve issues in the workflow team, such as a script for recovery of missing lumis or a diagnosis for requests that seem "stuck", or running for too long.

WMCore development

• I have been in charge of writing and editing the release notes for each release cycle of WMAgent, WorkQueue and ReqMgr (not ReqMgr2). The WMStats section is handled by Seangchan. https://twiki.cern.ch/twiki/bin/viewauth/CMS/DMWMReleaseNotes
• I usually go through the issues in github and organize them (https://github.com/dmwm/WMCore/issues?labels=&milestone=&page=1&state=open), tag them and put them in milestones according to priority and what has been discussed in the release planning meeting (https://twiki.cern.ch/twiki/bin/viewauth/CMS/DMWMReleasePlanning).
• Recently, I have been working on assigning development issues to Tapas and helping him in getting through them
• Present in the release planning meeting the recent developments for WMAgent and WorkQueue.
• Help in the Tier-0 development and testing which is lead by Dirk.

Tier-0 support (Before LS1)

• Run replays on demand by Offline and/or developer when there are changes in CMSSW, global tag or the WMAgent Tier-0 software.
• Monitor the LSF queue, the Tier-0 workflows by run and make sure that Express, Repack and PromptReco are running on schedule.
• Collect statistics and make weekly reports on the functioning of the system.