| Time |
Nick |
Message |
| 02:10 |
|
yar joined #evergreen |
| 06:57 |
|
agoben joined #evergreen |
| 07:07 |
|
rjackson_isl joined #evergreen |
| 07:26 |
|
collum joined #evergreen |
| 07:36 |
|
Dyrcona joined #evergreen |
| 08:05 |
|
rfrasur joined #evergreen |
| 08:36 |
|
collum joined #evergreen |
| 08:42 |
|
mdriscoll joined #evergreen |
| 08:49 |
|
bos20k joined #evergreen |
| 09:08 |
|
yboston joined #evergreen |
| 09:12 |
|
tlittle joined #evergreen |
| 09:58 |
|
aabbee joined #evergreen |
| 10:01 |
|
jeff joined #evergreen |
| 10:11 |
|
sandbergja joined #evergreen |
| 10:11 |
mdriscoll |
We have had several occasions of postgres crashing with out of memory errors. I probably need to tweak some settings in postgresql.conf, but the interesting thing is the searches that caused the oom. One was a really huge bunch of text, several others were obvious attemps at SQL injection. |
| 10:13 |
mdriscoll |
I fed one of the searches into a test server. It didn't run out of memory, but the search stayed running in the database even though the front end timed out. I had to restart postgres to get it to go away, pg_cancel did not work on it. Load average was high while this was running. |
| 10:22 |
jeff |
I would consider reproduceable searches like that a denial of service issue worth reporting as a private security bug. |
| 10:23 |
mdriscoll |
Jeff: thanks. Will do. |
| 10:23 |
jeff |
Interesting that pg_cancel_backend did not seem to have an effect. |
| 10:28 |
mdriscoll |
I couldn't shut down postgres either until I issued an immediate shutdown with pg_ctlcluster 9.4 main stop -m i |
| 10:35 |
jeff |
...which would have triggered a recovery on next start. Was that recovery successful, or did you run into any issues / scary warnings in the logs? |
| 10:37 |
jeff |
mdriscoll: did you first escalate to "fast" shutdown mode and then needed to use "immediate" because fast wasn't giving you results? |
| 10:41 |
|
jvwoolf joined #evergreen |
| 10:48 |
mdriscoll |
Jeff: I don't think I tried a fast shutdown. There weren't any issues with startup. |
| 10:57 |
|
Christineb joined #evergreen |
| 11:01 |
jeff |
mdriscoll: good to hear! |
| 11:03 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 11:25 |
|
sandbergja joined #evergreen |
| 11:56 |
|
stephengwills joined #evergreen |
| 12:06 |
|
jihpringle joined #evergreen |
| 12:40 |
|
khuckins joined #evergreen |
| 13:05 |
jeffdavis |
re: Tuesday's discussion of high open-ils.actor drone counts and setting lookups: bug 1848550 |
| 13:05 |
pinesol |
Launchpad bug 1848550 in Evergreen "Cache settings more aggressively in web client" [Undecided,New] https://launchpad.net/bugs/1848550 |
| 13:06 |
|
sandbergja_ joined #evergreen |
| 13:33 |
|
yboston joined #evergreen |
| 13:55 |
JBoyer |
jeffdavis++ |
| 14:44 |
|
sandbergja_ joined #evergreen |
| 15:49 |
Stompro |
terran++ thanks for testing my self check changes. |
| 16:22 |
|
jvwoolf left #evergreen |
| 16:30 |
|
yboston joined #evergreen |
| 16:53 |
|
khuckins joined #evergreen |
| 19:07 |
|
cmalm joined #evergreen |
| 22:37 |
|
sandbergja joined #evergreen |
| 23:01 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
