| Time |
Nick |
Message |
| 00:07 |
|
dbwells joined #evergreen |
| 02:38 |
|
dbwells joined #evergreen |
| 04:15 |
|
jamesrf joined #evergreen |
| 04:15 |
|
JBoyer joined #evergreen |
| 05:08 |
|
dbwells joined #evergreen |
| 06:02 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 07:27 |
|
rjackson_isl_hom joined #evergreen |
| 07:35 |
|
Dyrcona joined #evergreen |
| 07:38 |
|
dbwells joined #evergreen |
| 08:15 |
|
mantis1 joined #evergreen |
| 08:36 |
|
alynn26 joined #evergreen |
| 08:45 |
|
mmorgan joined #evergreen |
| 08:54 |
|
alynn26_away joined #evergreen |
| 08:58 |
|
rfrasur joined #evergreen |
| 08:59 |
Dyrcona |
JBoyer: So far the branch on Lp 1901760 is looking good for me with Chrome on Android. |
| 08:59 |
pinesol |
Launchpad bug 1901760 in Evergreen "Staff Client Whitescreen on login on all browsers on iOS/Android" [High,Confirmed] https://launchpad.net/bugs/1901760 - Assigned to Jason Stephenson (jstephenson) |
| 09:00 |
JBoyer |
Dyrcona++ |
| 09:08 |
Dyrcona |
Works on my daughter's iPad, too, but I noticed the hamburger menu options appear greyed out or at least very light, though they work. Must be a style. |
| 09:08 |
csharp |
seeing an issue only on our production server where the search box on the splash screen for the staff client is apparently double-URL-encoding spaces and creating things that look like "query=harry%2520potter" and showing "harry%20potter" in the search box in the staff catalog |
| 09:09 |
csharp |
I've chased things as far as I know how and can't see what's causing it - I'm just stumped and have been for a few days now |
| 09:09 |
csharp |
on the other hand, this appears to be the highest-priority issue we have right now after a major upgrade, so there's that :-) |
| 09:10 |
csharp |
our test versions running the same code are *not* experiencing the problem and encoding the URL params correctly |
| 09:11 |
Dyrcona |
All O/S and package updates applied to both sets of machines? |
| 09:11 |
csharp |
yes |
| 09:11 |
csharp |
and another factor is that production is on Ubuntu 16.04 and our test machines have been on 18.04 |
| 09:12 |
csharp |
I went down a rabbit hole fighting with a local VM trying to install OpenSRF last night and haven't resumed |
| 09:12 |
csharp |
(on 16.04) |
| 09:13 |
Dyrcona |
I guess something recent doesn't like 16.04, but tbh, I've seen weird behavior that just "goes away" after a clean install. (I assume this after an in place upgrade on production.) |
| 09:27 |
|
dbwells joined #evergreen |
| 09:28 |
Bmagic |
csharp: strange indeed |
| 09:29 |
csharp |
Dyrcona: yes - it's an in-place upgrade and we're moving to fresh 18.04 installs very soon |
| 09:30 |
Bmagic |
Though, I just wasted almost 2 days working a problem similar. And finally, I decided to setup a whole new test environment, fresh install DB, fresh install Application server, but otherwise identical - and the problem was no longer there! Which was a bad thing, because I couldn't diagnose it |
| 09:30 |
csharp |
so you're confirming my working theory - gonna keep going on my 16.04 test install |
| 09:31 |
csharp |
it may come down to one of those things where my cautious "don't upgrade too many things at once" strategy means I have problems I wouldn't have seen if I had gone ahead and upgraded OS along with :-/ |
| 09:31 |
Bmagic |
Right now, the only thing I could think of was memcached on production is caching queries and introducing the difference |
| 09:31 |
|
jvwoolf joined #evergreen |
| 09:31 |
csharp |
huh - that's interesting |
| 09:32 |
Bmagic |
but that's a real long shot, because, in theory, memcached would expire the cache in some kind of reasonable amount of time |
| 09:32 |
* Dyrcona |
suspects a Perl module. |
| 09:34 |
csharp |
yeah - that makes sense, Dyrcona |
| 09:34 |
csharp |
Bmagic: yeah, I was thinking the same |
| 09:38 |
csharp |
apparmor-- |
| 09:38 |
csharp |
that was my rabbit hole, now I can continue |
| 09:41 |
Bmagic |
rabbit_hole++ |
| 09:46 |
Dyrcona |
So, during the previous developer meeting, it was suggested that we have a freeze on pushing to the repos starting on Monday, for a Wednesday release: http://evergreen-ils.org/meetings/evergreen/2021/evergreen.2021-01-05-15.05.log.html # Search for "freeze." It is the second result. |
| 09:47 |
Dyrcona |
Is that something that we still want to do, and if so, should someone send an email about it today? I was considering starting the branches on Monday. |
| 09:52 |
|
stephengwills joined #evergreen |
| 09:53 |
berick |
csharp: in case it hasn't come up yet, bug 1892435 |
| 09:53 |
pinesol |
Launchpad bug 1892435 in Evergreen "Splash page catalog search doesn't work for multi-word queries" [Undecided,New] https://launchpad.net/bugs/1892435 |
| 10:03 |
gmcharlt |
Dyrcona: and CC JBoyer: TBH, given pending stuff we know we want in the next maintenance release - and the general tuit crash of January 2021 - perhaps we should consider pushing next release to either 2/3 or 2/10 |
| 10:03 |
gmcharlt |
(& CC et al) |
| 10:04 |
JBoyer |
I wouldn't argue with that at all. |
| 10:12 |
* mmorgan |
consults Launchpad and sees 46 signedoff bugs targeted for 3.6.2 |
| 10:15 |
Dyrcona |
The tuits crashed months ago, TBH. |
| 10:17 |
Dyrcona |
The reason for the freeze is that I was planning to push 2 or 3 security patches, which requires more coordination than usual. Pushing the releases back another week makes sense under the circumstances. |
| 10:24 |
* mmorgan |
would be in favor of pushing the releases ahead if it means more fixes get in. |
| 10:29 |
|
tlittle joined #evergreen |
| 10:40 |
Dyrcona |
Re: https://bugs.launchpad.net/evergreen/+bug/1892435/comments/12 A reason that this may not happen with Firefox is that + is also used to replace ' ', instead of %20, so that could explain why it may not happen with one browser, but happens with another. |
| 10:40 |
pinesol |
Launchpad bug 1892435 in Evergreen "Splash page catalog search doesn't work for multi-word queries" [Undecided,New] |
| 11:15 |
csharp |
berick: thanks! |
| 11:15 |
csharp |
berick++ |
| 11:16 |
csharp |
TFW you try to increment but are pressing Ctrl rather than Shift and the magnification increases |
| 11:17 |
Dyrcona |
:) |
| 11:21 |
Dyrcona |
Looking in our Makefile.ams gives me the heebie-jeebies. |
| 11:34 |
Dyrcona |
Hmm... I can't find the code that installs the eg2 files. That said, I think I see what happens just looking at the files. |
| 11:35 |
berick |
Dyrcona: i think it's part of the generic whole-web-dir copy |
| 11:36 |
Dyrcona |
berick: That's what it looks like, yeah. |
| 11:38 |
csharp |
rebuilding npm/node_modules everywhere does not solve the problem for us :-( |
| 11:38 |
Dyrcona |
The build happens in Open-ILS/web/eg2/en-US during ng build. # For the log/myself thinking out loud. |
| 11:40 |
Dyrcona |
csharp: What happens if you follow nfburton's suggestion in https://bugs.launchpad.net/evergreen/+bug/1892435/comments/11 ? |
| 11:40 |
pinesol |
Launchpad bug 1892435 in Evergreen "Splash page catalog search doesn't work for multi-word queries" [Undecided,Confirmed] |
| 11:41 |
Dyrcona |
It's sounding like a bug in our Perl code, though. Something encoding an already encoded string. |
| 11:42 |
Dyrcona |
Anyway, I'm looking into just copying the Angular (eg2) and AngularJS staff client files because I want to patch them without having to do make install. |
| 11:43 |
Dyrcona |
Looks like just copying the appropriate directories will do it. |
| 11:43 |
csharp |
I'm expecting to find that it happens on Ubuntu 16.04 and not 18.04 and that bolsters the Perl theory since it's pretty much the only difference in that stack |
| 11:44 |
berick |
Dyrcona: can confirm copying files into place is fine |
| 11:44 |
csharp |
I haven't installed manually in a while - I used to have the step memorized :-) |
| 11:44 |
Dyrcona |
If it really is on 16.04 and not 18.04, then it's likely not our code, but some module we use. |
| 11:44 |
* berick |
pats ansible on the head |
| 11:44 |
Dyrcona |
:) |
| 11:44 |
csharp |
yes, I have a script that builds eg2 on an app server and rsyncs it to the other apps |
| 11:45 |
csharp |
rsyncs Open-ILS/web/eg2 to /openils/var/web/eg2 to be clear |
| 11:45 |
Dyrcona |
I install manually and use cssh to do machines in batches. |
| 11:46 |
Dyrcona |
Or just build new "model" VMs and clone/copy those. |
| 11:51 |
|
sandbergja joined #evergreen |
| 12:05 |
|
jvwoolf1 joined #evergreen |
| 12:05 |
Dyrcona |
Bleh. Vendors using hashes in URLs complicates my record management code. |
| 12:06 |
|
jihpringle joined #evergreen |
| 12:20 |
JBoyer |
jvwoolf1++ |
| 12:21 |
jvwoolf1 |
redtape-- |
| 12:29 |
csharp |
I think the staff catalog may be introducing new proliferation of actor drones |
| 12:30 |
csharp |
berick's fixes for bug 1896285 had us doing just fine before the upgrade, now were back to drone exhaustion |
| 12:30 |
pinesol |
Launchpad bug 1896285 in Evergreen 3.5 "Use batch methods for multi-row grid actions" [Undecided,New] https://launchpad.net/bugs/1896285 |
| 12:34 |
mmorgan |
csharp: We have been seeing the same thing on 3.6, we also applied those fixes, but are still seeing the problem. |
| 12:35 |
csharp |
mmorgan: good to confirm - thanks |
| 13:11 |
berick |
if anyone can track it down i can help fix |
| 13:24 |
pinesol |
[evergreen|Bill Erickson] LP1901760 Improve SharedWorker non-support handling (AngJS) - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=3c1188f> |
| 13:24 |
pinesol |
[evergreen|Bill Erickson] LP1901760 Improve SharedWorker non-support handling (Angular) - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=c9f731f> |
| 13:24 |
pinesol |
[evergreen|Bill Erickson] LP1901760 Remove SharedWorker testing cruft - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=66e7e45> |
| 13:27 |
Dyrcona |
Y'kinow with that fix that I pushed applied, your users on Android Chrome and iOS devices will not be caching settings and will not have the benefit of the fix for bug 1848550. All it takes is 1 errant client to max out the actor drones. |
| 13:27 |
pinesol |
Launchpad bug 1848550 in Evergreen "Cache settings more aggressively in web client" [Undecided,Fix released] https://launchpad.net/bugs/1848550 |
| 13:29 |
berick |
ok, yeah, it's time for a comprehensive fix. i'll open a bug to support max-parallel-requests on the client side. |
| 13:31 |
csharp |
confirmed broken splash page search box on a fresh install on 16.04 (of PINES-customized 3.6.1) |
| 13:32 |
csharp |
meanwhile, terranm has worked up a patch to redirect to the eg2 splash page, as nfBurton28 mentioned in his bug comment |
| 13:33 |
csharp |
but I'm nervous about it because I got caught in a eg/eg2 workstation registration loop where eg2 didn't recognize my workstation and eg1 didn't recognize the new eg2 workstation (or so it seemed) |
| 14:50 |
|
Cocopuff2018 joined #evergreen |
| 15:43 |
|
mantis1 left #evergreen |
| 15:51 |
csharp |
I'll be damned if I can find where the "double encoding" happens in this - it appears to enter the eg2 catalog with the %20 intact as an actual search character |
| 15:52 |
Dyrcona |
Well, ASCII 37, i.e. 0x25, is %, so that looks like what is happening. |
| 15:53 |
csharp |
yeah |
| 16:15 |
berick |
csharp: https://bugs.launchpad.net/opensrf/+bug/1912834 |
| 16:15 |
pinesol |
Launchpad bug 1912834 in OpenSRF "Browser client should limit the number of parallel requests" [High,New] |
| 16:15 |
berick |
has a pullrequest now |
| 16:15 |
csharp |
berick: awesome! |
| 16:31 |
|
jweston joined #evergreen |
| 16:42 |
csharp |
berick: getting Uncaught TypeError: this.sendFromQueue is not a function |
| 16:42 |
csharp |
at WebSocket.socket.onopen (opensrf_ws.js:87) |
| 16:42 |
csharp |
this is applying to opensrf 3.2.1/Evergreen 3.6.1 on Ubuntu 16.04 |
| 16:42 |
csharp |
sorry 18.04 |
| 16:43 |
|
stephengwills left #evergreen |
| 16:45 |
csharp |
gonna have to leave it for now though, so no rush |
| 16:46 |
csharp |
for anyone curious, I just used decodeURIComponent() on the query string to get the angjs staff splash search box sorta working |
| 16:46 |
csharp |
https://git.evergreen-ils.org/?p=evergreen/pines.git;a=commit;h=3c9085df254bcd45d03c8e009bca5ef3bc2f6d0e |
| 16:46 |
csharp |
the search now works, but %20 still appears where the space should be in the search bar and %2520 in the URL |
| 16:47 |
csharp |
but it's enough for us |
| 16:47 |
csharp |
especially when the better option is "click literally 1/2 inch downwards, then type in your search" |
| 16:48 |
csharp |
ok, hanging it up - hope you enjoyed my closing monologue |
| 16:48 |
csharp |
@monologue |
| 16:48 |
pinesol |
csharp: Your current monologue is at least 13 lines long. |
| 17:00 |
berick |
csharp: arg, that branch has 2 commits, the first one is old/bogus. |
| 17:00 |
* berick |
cleans it up |
| 17:02 |
berick |
force-pushed and fixed |
| 17:26 |
|
mmorgan left #evergreen |
| 18:02 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 18:22 |
|
sandbergja joined #evergreen |
| 18:40 |
|
jvwoolf joined #evergreen |
| 18:47 |
|
jihpringle joined #evergreen |
| 20:03 |
|
mrisher joined #evergreen |
| 22:33 |
|
sandbergja joined #evergreen |
