Participants
============
1. Brett
2. Hossein
3. Cloph
4. guilhem
Agenda
======
* Hossein: Is the number of commits reported in the Weblate and Kibana
dashboard correct?
+ Hundreds of changes are being shown as thousands
+ The number of edits is approximately multiplied by the number of
branches, because many small strings exist in multiple branches
+ The number can be filtered by a branch.
+ In this way, the number can be considered OK.
+ guilhem: Check the “Data Status” dashboard to see when the metrics were
last imported
* Hossein: Is the number of git commits reported in the Kibana dashboard
correct?
+ The author should be found using the "Add filter" method then adding the
author_name, and not by just searching for the name in the lucene input
field. (Hint: Then the filter can be pinned, and used with other pages
like Git, Bugzilla, etc.)
+ In this way, the reported number matches the count in the default branch.
+ A simple statistics page is created for MC (Guilhem)
- The statistics is pulled from Kibana dashboard JSON API and is
displayed as a simple web page
- Should be possible in possible to show the output for everyone in
his/her SSO profile page
- Hossein: There are badges that are awarded to the people in the "Month
of LibreOffice campaign"
.
https://wiki.documentfoundation.org/Marketing/Month_of_LibreOffice#Badges
. Because the data is available in the dashboard, it is possible to
give people badges in their sso profile page
. We should ask Mike for more information on criterias to give badges
* Postgres DB backups (Brett)
+ I discovered that pg_basebackup does not feasibly work in a push mechanism,
only pull.
- If we are to go the pg_basebackup + remote WAL archiving route, we'll
need to either switch the WAL archives to a pull mode (a little more
complicated to set up) or set up both push *and* pull (push for WAL,
pull for pg_basebackup).
+ G: We want backups that don't require full dumps frequently (hundreds of
GiB is a pain to have to transfer often). Custom logic/scripting is not
desired, let's see what other people do (we're not the only ones with
this issue).
+ Brett: there is https://pgbackrest.org/ also, need to check in more
details - natively packaged in Debian and is a popular backup solution.
. Documentation looks much better than barman's, we can follow their
recommendations to limit any issues that would be affect unique setups
+ G: Will upgrade the backup box/etherpad to Debian 11 to have a homogenous
baseline (edit: done since the call)
* Jenkins hiccups:
+ occasionally runs out of heap space and refuses new connections (Jenkins
bug?)
+ edit: possibly caused by an issue deeper in the virtualizion stack, now
mitigiated/fixed
+ callgrind job DoS CI and filling disks
- g: not really sustainable at the current rate
- cloph: the output is now compressed immediately after it is finished
(~250 MB compressed / 3GB uncompressed)
- currently keeping 15 days of logs, can reduce to 7 days if needs be
+ The maximum number of concurrent jobs is now limited to avoid problems
- One instance of such a problem: Changing many submissions from "Work in
progress" to "Ready for review"
* Systems upgrades (g): upgraded a dozens systems last week, notably AskLibO
and Nextcloud
* Next call: Dec 21 at 17:30UTC
--
Guilhem.
--
To unsubscribe e-mail to: [email protected]
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/
Privacy Policy: https://www.documentfoundation.org/privacy
