Skip to content

postgresql: Increase some timeouts to get origin visits

Antoine Lambert requested to merge generated-differential-D8124-source into generated-differential-D8124-target

Even if missing index to speedup origin visit queries has been added to replica database, the configured timeouts for origin_visit_get_with_statuses and origin_visit_find_by_date were still too low to avoid query timeouts in production.

After performing some tests locally, bumping them to 2000ms makes the timeouts go away.

Related to swh/infra/sysadm-environment#4386

Should fix SWH-STORAGE-1C0, SWH-STORAGE-1DM and SWH-STORAGE-1DJ.

Test Plan

I created a SSH tunnel to be able to talk to production database server from my machine connected to the VPN.

(swh) ✔ ~/swh/swh-environment/swh-web [master ↑·1|✚ 1⚑ 143] 
16:26 $ ssh -L 8900:192.168.100.103:5433 somerset.internal.softwareheritage.org
Linux somerset 5.15.35-3-pve #1 SMP PVE 5.15.35-6 (Fri, 17 Jun 2022 13:42:35 +0200) x86_64

1 updates could not be installed automatically. For more details,
see /var/log/unattended-upgrades/unattended-upgrades.log

The programs included with the Debian GNU/Linux system are free software;
the exact distribution terms for each program are described in the
individual files in /usr/share/doc/*/copyright.

Debian GNU/Linux comes with ABSOLUTELY NO WARRANTY, to the extent
permitted by applicable law.
Last login: Wed Jul 13 14:25:59 2022 from 192.168.101.8
anlambert@somerset:~$

Then I executed the storage RPC server locally with the following command and configuration:

swh storage -C ~/.config/swh/storage.yml rpc-serve
storage:
  cls: local
  db: "host=localhost port=8900 dbname=softwareheritage user=guest"
  
  objstorage:
    cls: remote
    url: http://objstorage.internal.softwareheritage.org:5003 # does not work but fortunately not needed for my tests

I updated swh-web configuration to use my local storage RPC server:

storage:
  cls: remote
  url: http://localhost:5002/

...

Finally, I tweaked the timeout value for the origin_visit_get_with_statuses and origin_visit_find_by_date methods until finding one that does not make queries timeout, 2000ms seems to be enough.

Migrated from D8124 (view on Phabricator)

Merge request reports