Choose timezone
Your profile timezone:
BiLD-DevOps
→
Europe/Zurich
2-R-014 (CERN)
2-R-014
CERN
Description
Bi-Weekly LHCbDirac DevOps meeting
Present: Chris,Vladimir, Federico, Joel, Philippe, Zoltan, Alexander
Remote: Marcelo
Minutes/matters from last meeting
gitlab-ci: Runs pytest, pylint on changed line and coverage.
Vobox root login: still not clear why Federico can login as root. To be added to the list: Zoltan, Philippe, Federico.
runsv (https://github.com/DIRACGrid/DIRAC/issues/3393): all the machines that have been rebooted use the system runsv. The others will at the next restart
HTCondor: we do not suffer anymore from proxy lifetime problem at CERN however there are still aborted pilots and still lazarus pilots.
Backup of the CS (https://its.cern.ch/jira/browse/LHCBDIRAC-742): done on EOS. Based on alias to choose the machine to backup
dirac.cfg and bashrc on puppet (https://its.cern.ch/jira/browse/LHCBDIRAC-696): done for all the machines that are the same. The configuration of the machines for the webportal (lbvobox33), the logse(volhcb04), the main CS (lbvobox201) and boinc (lbboinc01,lbvobox46) are handled by hand.
LHCbDIRAC installation: Status of servers, DBs, build machines…
MySQL:
The last few issues we had with MySQL 5.7 (Production Request DB and Accounting) seem to be fixed.
We would like to schedule the upgrade of the production DB as well as the change of password for next week.
Chris to check with DBOD how is the upgrade happening (duration, downtime, etc).
LogSE:
https://its.cern.ch/jira/browse/LHCBDIRAC-350
Made all the discussion based on the assumption that the MCStat Tools do not exist.
Currently, the log is uploaded in different formats depending on the situation (standard case, files bigger than 1M, failover, etc). They are uploaded to Log-SE which is a Dirac SE, and run its own webserver.
It was agreed:
* to always tar everything in one archive
* we will use the EOS https server
* the TarGZAgent will disappear, we will use "eos archive" in the TransformationCleaningAgent when archiving a transformation
We need to check a few things:
* The URL set in the job parameters should be updated
* We should check how the web portal gets the link
* Which operation should be used for the failover
This needs to be presented to Gloria.
DIRAC/LHCbDIRAC Releases planning and installation
Echo:
The move to ECHO has many impacts both operationally and from the dev point of view
All the tasks are here: https://its.cern.ch/jira/browse/LHCBDIRAC-784
From the operations point of view, the idea is to let RAL do the move of the physical files themselves.
We will use a xroot redirector to read, regardless of whether it is on the CEPH or the Castor storage.
We will write using GSIFTP directly to Echo.
The main issue with this is that there are no third party copy with xroot to the dCache sites.
Conferences planning and reports
All people intending in presenting an abstract to CHEP should post it here: https://indico.cern.ch/event/680068/
Currently, Federico plans on going, Philippe (talk about Operational aspects ? ), Joel.
Maybe Zoltan can make an abstract about BKK optimization and uses of ElasticSearch
There are minutes attached to this event.
Show them.