staging is overloaded
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Launchpad itself |
Fix Released
|
High
|
[LEGACY] Canonical WebOps |
Bug Description
Staging is being routinely overloaded:
https:/
Some causes are the check-teamparti
13:46 < spm> 15900 launchpa 20 0 388m 126m 8180 R 98 3.3 0:05.00 /usr/bin/python /srv/staging.
13:46 < spm> 15901 bzrsyncd 20 0 279m 88m 6284 R 24 2.3 0:03.55 /usr/bin/python /srv/bzrsyncd.
13:46 < spm> 15914 bzrsyncd 20 0 350m 101m 7968 R 24 2.6 0:04.01 /usr/bin/python /srv/bzrsyncd.
13:46 < spm> 15906 bzrsyncd 20 0 279m 89m 6284 R 23 2.3 0:03.28 /usr/bin/python /srv/bzrsyncd.
13:46 < spm> 15929 launchpa 20 0 376m 118m 8092 R 22 3.1 0:05.06 /usr/bin/python /srv/staging.
13:46 < spm> 9052 launchpa 20 0 748m 392m 3440 R 7 10.2 22:35.15 /usr/bin/python -S ./foaf-
(the column with 98 on the top is CPU)
We can't QA performance changes to the appservers effectively while we're overloading the system: some noise is expected, but saturated throws a more than 100% delay into things, which makes ok things timeout and as a knockon effect can waste a lot of time analysing perfectly ok code.
We'll probably want to add RT's to fix this all up, and perhaps we need a separate box for these tasks vs the staging appserver instance(s)
check-teamparti cipation: bug 633648