Time |
Nick |
Message |
07:28 |
|
rjackson_isl_hom joined #evergreen |
07:36 |
|
collum joined #evergreen |
07:42 |
|
kworstell-isl joined #evergreen |
07:54 |
|
rjackson_isl_hom joined #evergreen |
07:54 |
|
BDorsey joined #evergreen |
07:54 |
|
abneiman joined #evergreen |
07:54 |
|
jweston joined #evergreen |
07:54 |
|
scottangel joined #evergreen |
07:54 |
|
DaMobi joined #evergreen |
07:54 |
|
jeff joined #evergreen |
07:54 |
|
JBoyer joined #evergreen |
07:54 |
|
troy joined #evergreen |
07:54 |
|
book`_ joined #evergreen |
07:54 |
|
bshum_ joined #evergreen |
07:54 |
|
dluch_ joined #evergreen |
07:54 |
|
Bmagic joined #evergreen |
07:58 |
|
rfrasur joined #evergreen |
08:00 |
|
BDorsey joined #evergreen |
08:00 |
|
abneiman joined #evergreen |
08:00 |
|
jweston joined #evergreen |
08:22 |
|
mantis1 joined #evergreen |
08:48 |
|
mmorgan joined #evergreen |
08:56 |
|
Dyrcona joined #evergreen |
08:57 |
|
jonadab joined #evergreen |
10:33 |
|
dguarrac joined #evergreen |
10:49 |
|
stephengwills joined #evergreen |
12:28 |
|
mantis1 joined #evergreen |
12:42 |
|
collum joined #evergreen |
13:27 |
|
rjackson_isl_hom joined #evergreen |
13:34 |
jeffdavis |
just had 21400 of these in a 1-minute period: open-ils.actor.user.has_work_perm_at.batch null, ["VIEW_USER"] |
13:52 |
jvwoolf |
jeffdavis: I've seen that before, but I can't recall if we were able to figure out where they originated from |
13:58 |
jeffdavis |
looking at the transaction IDs(?) in the logs in sequence, it appears to go: (1) attempt to retrieve an expired authtoken, (2) delete the associated auth session, (3) spam the servers with those open-ils.actor calls for VIEW_USER with null authtoken |
13:59 |
jeffdavis |
I don't recall seeing this on EG 3.7. Looking at where VIEW_USER perms are checked in the client, looks like this is probably an effect of changes made to support consolidated patron messages. |
14:01 |
jeffdavis |
I wonder if this is another case where a session timeout results in a flood of requests, similar to bug 1940698 |
14:01 |
pinesol |
Launchpad bug 1940698 in Evergreen "Duplicate open-ils.search.z3950.search_class calls lead to drone exhaustion" [Undecided,New] https://launchpad.net/bugs/1940698 |
14:03 |
jeffdavis |
we've seen this at least once before, at a time where the only other activity was someone's un-logged-out client session redirecting to the login page in the middle of the night |
14:11 |
Dyrcona |
It could be. Someone had an idea that a session timeout sometimes leads to the client spamming its "last" request. |
14:12 |
Dyrcona |
It might be in that bug you mentioned. |
14:17 |
Dyrcona |
Yeah, it was gmcharlt. |
14:17 |
Dyrcona |
Maybe the consolidated messages interface is doing it, too? |
14:21 |
jeffdavis |
I don't think it's the messages UI, I think it's changes to the patron app that were made as part of that work |
14:21 |
jeffdavis |
reported to Launchpad as bug 1990306 |
14:21 |
pinesol |
Launchpad bug 1990306 in Evergreen "Flood of open-ils.actor.user.has_work_perm_at.batch requests for VIEW_USER with null authtoken" [Undecided,New] https://launchpad.net/bugs/1990306 |
14:22 |
Dyrcona |
OK. I misunderstood. |
14:34 |
Dyrcona |
Huh... Wonder what I did that I had to reset my bash terminal? |
14:35 |
Dyrcona |
Anyway, I wonder if the changes for the messaging center hold any clues to the problem jeffdavis mentioned? |
14:36 |
jeffdavis |
I specifically suspect that the VIEW_USER lookup that now happens in the resolve portion of egPatronApp is responsible, but don't know Angular well enough to really address it |
14:36 |
jeffdavis |
it's not the messaging center per se, just an unexpected consequence of some refactoring (if I'm right |
14:37 |
Dyrcona |
Yeah. |
14:38 |
Dyrcona |
jeffdavis: Do you have commit abf033c19050af93973a2d204378918cc6253749? |
14:39 |
Dyrcona |
or 3bc23d8fcf6c5bf0666cb204d3c4b009db562476 rather? |
14:39 |
jeffdavis |
yes, we've got that one |
14:41 |
jeff |
I thought that we had seen "sleeping/inactive client spams server" with open-ils.auth.session.retrieve on 3.7, but I might be wrong. Have to check. |
14:42 |
jeff |
(on wake / becoming active again) |
14:43 |
jeff |
I think at the time we had a theory about it being the background tabs looping on fresh login, but I don't recall why. |
14:43 |
jeff |
Also, might be a different-but-similar circumstance. |
14:43 |
jeff |
especially because I'm talking 3.7 and jeffdavis isn't, I'm pretty sure. |
14:43 |
jeffdavis |
yeah we're 3.9 now |
14:43 |
Dyrcona |
I was thinking that settimeout might be a problem, but it probably isn't. I noticed a commit that removed a refreshpage(): 4460554d754a08023425c6d476ac8293f6af3b29 You probably have that one if you have the other, though. |
14:47 |
Dyrcona |
jeffdavis: Do you suspect ac037f5143b33fdb03fbaa445b82a274f1b1efc1 ? The commit hash is from rel_3_9, not master. |
14:48 |
Dyrcona |
It adds the VIEW_USER perm check to eg.startup.go() with a nice FIXME comment. Maybe following though on the FIXME would help? |
14:49 |
jeffdavis |
yes, that's the bit my eye is on |
14:49 |
Dyrcona |
jeffdavis++ |
14:49 |
Dyrcona |
Is it possible to delete a promise? |
14:52 |
Dyrcona |
@who always keeps promises |
14:52 |
pinesol |
Stompro always keeps promises. |
15:00 |
|
Christineb joined #evergreen |
15:18 |
|
mmorgan1 joined #evergreen |
16:08 |
|
Guest4 joined #evergreen |
16:37 |
|
mmorgan joined #evergreen |
17:17 |
|
stephengwills left #evergreen |
17:20 |
|
mmorgan left #evergreen |
18:14 |
|
mantis1 joined #evergreen |