archive: Increase the number of backend workers
gunicorn-swh-webapp service has currently 32 workers. Cranking up from 4 to 16 will likely decrease the socket timeout occurrences [1] when a storage worker is taking longer than anticipated to give back its computation result.
- [1]
10:22 <swhbot> icinga RECOVERY: service swh-storage api (localhost) on moma.softwareheritage.org is OK: HTTP OK: HTTP/1.1 200 OK - 482 bytes in 0.001 second response time
10:23 <swhbot> icinga RECOVERY: service swh webapp content known on moma.softwareheritage.org is OK: HTTP OK: HTTP/1.1 200 OK - 530 bytes in 0.025 second response time
10:23 <swhbot> icinga RECOVERY: service swh webapp revision log end to end on moma.softwareheritage.org is OK: HTTP OK: HTTP/1.1 200 OK - 110006 bytes in 0.200 second response time
Test Plan
octocatalog-diff
diff origin/production/moma.softwareheritage.org current/moma.softwareheritage.org
*******************************************
File[/etc/gunicorn/instances/swh-storage.cfg] =>
parameters =>
content =>
@@ -41,5 +41,5 @@
# custom settings
bind = "unix:/run/gunicorn/swh-storage/gunicorn.sock"
-workers = 4
+workers = 16
worker_class = "sync"
timeout = 3600
*******************************************
Gunicorn::Instance[swh-storage] =>
parameters =>
settings =>
workers =>
- 4
+ 16
*******************************************
*** End octocatalog-diff on moma.softwareheritage.org
Migrated from D4725 (view on Phabricator)