Minutes from the Tue Apr 17 infra call

classic Classic list List threaded Threaded
1 message Options
Guilhem Moulin Guilhem Moulin
Reply | Threaded
Open this post in threaded view

Minutes from the Tue Apr 17 infra call


  1. guilhem
  2. Dennis
  3. Christian
  4. rajudev [libregeekingkid] (in IRC)
  5. Brett


 * [Florian Reisinger (unable to participate) reisi007]
   Can this folder [ https://dev-builds.libreoffice.org/daily/ ] be cleaned
   + We can improve the cronjob to remove empty dirs & broken symlinks, for
     now we can do it manually
   + AI guilhem: expose 3 JSON dirs on download.tdf, downloadarchive.tdf and
     dev-builds with the directory structure
 * Monitoring
   + Grafana and prometheus deployed on https://monitoring.documentfoundation.org
   + Do we want to install the node exporter on each VM?  All recent VMs
     expose their state to the host through QEMU guest agent
     https://wiki.qemu.org/Features/GuestAgent and we could use that instead
     - Brett: metrics aren't necessarily accurate when measured from the
     - We might have to write our own exporter though
   + Alerting: do we want that in prometheus, or on the Grafana panels? (so
     far only graph panels support that, but it should be possible with
     singlestats and table panels in later versions)
     - Brett: prometheus is nice for version controlling (plain YAML files).
       prometheus' altert manager is feature rich
   + collectd: OK to remove from salt core, and uninstall from the VMs?
     - OK (keep it on the hypervisors for now)
   + Exporters to look for, test and adopt:
     - libvirtd (so we see whether we overcommit RAM or vCPUs)
     - gluster: throughput, latency, healing stats
     - postfix: mail queue
     - RDBMS: slow queries, transactions
     - nginx: cache hits/misses, requests count, method, status code
   + Would be nice to expose the metrics from the blackbox exporter (or
     associated dashboards), not sure if that's feasible without exposing the
     whole prometheus data source.
     - Brett: Snapshots is the only possibility currently
 * Bugzilla upgrade: currently hosted our last prod Debian 7 VM (codename
   Wheezy), EOL on May 31.  Trash it and replace it with a fresh Debian 9 one
   from the new baseline?
   + upgrade vm144 (test bugzilla) to Stretch and test it there first
 * Upgrade VMs and hypers to Stretch in late Q2 / early Q3, starting with Charly
   - Upgrade to MW1.30 (https://www.mediawiki.org/wiki/LTS) - released Dec
     2017; MW1.29 EOL July 2018
   - MW1.31 is LTS but not released yet, need to upgrade to MW.30 first
   - Dennis would like to have fun and do server side stuff again --> test-wiki
   - Dennis would love to test afterwards to import content from TRWIKI (stuff
     cleaned up from http://eskiwiki.libreoffice.org.tr and
   - First Upgrade MW; afterwards test new PHP7.0 from Debian 9 on TESTWIKI
 * SSO adoption:
   + 32/190 (16%) TDF members not in LDAP, after two individual reminders
   + 51/176 (29%) recent wiki editors not in LDAP, after a banner and one
     individual reminder; Dennis: SQL query seems right:

         SELECT DISTINCT user_email
               FROM user JOIN recentchanges ON user_id = rc_user
              WHERE rc_source = "mw.edit"
                AND rc_bot = 0
                AND rc_deleted = 0
                AND user_email_authenticated IS NOT NULL;

     agreed to switch auth to SAML during the last (March) infra call since
     the %-age is <33%, didn't get around to it yet
   + Cloph: need to have some public instructions for LDAP DIT manipulation
 * Next call: Tuesday May 15 2018 at 18:30 Berlin time (16:30 UTC)


To unsubscribe e-mail to: [hidden email]
Problems? https://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: https://wiki.documentfoundation.org/Netiquette
List archive: https://listarchives.libreoffice.org/global/website/
All messages sent to this list will be publicly archived and cannot be deleted