Time |
Nick |
Message |
03:09 |
|
book` joined #evergreen |
07:07 |
|
kworstell-isl joined #evergreen |
07:45 |
|
BDorsey joined #evergreen |
08:40 |
JBoyer |
Stompro++ # Catching up from yesterday |
08:41 |
Stompro |
gmcharlt++ and jeffdavis++ for the code that I copied to get it working. |
09:05 |
|
Dyrcona joined #evergreen |
09:15 |
|
dguarrac joined #evergreen |
09:27 |
Bmagic |
kmlussier++ |
09:50 |
|
terranm joined #evergreen |
10:06 |
|
sandbergja joined #evergreen |
10:08 |
sandbergja |
Hey friends, I was thinking of cutting a 3.10.2 release over the weekend to get some bugfixes out the door and for my own learning. Don't want to interfere with any other release plans, tho. |
10:08 |
sandbergja |
any objections? |
10:11 |
Dyrcona |
I don't object, but I want to put out that releases are scheduled for next Wednesday though I'm not aware of any discussion about doing them next week. |
10:13 |
Dyrcona |
s/put/point/ |
10:18 |
|
kworstell_isl joined #evergreen |
10:19 |
Bmagic |
sandbergja++ # what about the other EG lines? or are you thinking just 10.x? |
10:20 |
sandbergja |
Dyrcona: that's a good point. I could volunteer to do it Wednesday evening instead -- just as long as it is outside of work hours. |
10:22 |
sandbergja |
Bmagic: I just had my eyes on 3.10 for now, but don't want to preclude anyone from working on the others, of course |
10:24 |
sandbergja |
...but at the same, I don't think I should volunteer if we're going to do security releases. Dropping a security patch outside of work hours just seems like a bad thing. |
10:31 |
Dyrcona |
I should find out if we're doing releases this Wednesday or not. We have slipped a good bit over the past few years. |
10:34 |
sandbergja |
Dyrcona++ |
10:35 |
sandbergja |
releases are hard |
10:40 |
Dyrcona |
It's the coordination that gets tricky. |
10:44 |
Dyrcona |
Release notes are also a hold up sometimes. |
10:44 |
Dyrcona |
I just emailed the dev list to see what others think. |
10:45 |
Dyrcona |
There is also the #evergreen-release channel here on libera for more discussion of release related issues. |
10:59 |
pinesol |
News from commits: LP#1778567: don't return cached list/tree before updating <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=25d666d7977df5a277c8548ded2ac3ab27e4fdec> |
10:59 |
pinesol |
News from commits: LP2002363 Aria labels for catalog search +/- buttons <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=bec822944c5c030c87851dac27c6fef4d693a25c> |
11:06 |
csharp_ |
Dyrcona++ |
11:21 |
sandbergja |
Dyrcona++ |
12:14 |
|
jihpringle joined #evergreen |
12:24 |
|
kmlussier joined #evergreen |
12:28 |
terranm |
When refreshing a test server, it's really helpful to log into the correct one. |
12:39 |
Dyrcona |
terranm++ |
12:40 |
Dyrcona |
sandbergja++ |
12:40 |
Dyrcona |
It looks like we'll have point releases next week, and there is a high likelihood of there being one or more security patches in the releases. |
12:46 |
csharp_ |
for when you do an OS upgrade and it breaks NPM all to hell: https://pastebin.com/aWcn6jkC |
12:46 |
csharp_ |
(thanks to gmcharlt for the kernel of this back at the 2019 hackaway) |
12:57 |
Dyrcona |
csharp_++ I have a less comprehensive version of that, but I think I'll steal yours. |
12:57 |
Dyrcona |
I find it useful sometimes to wipe out Node and reinstall it when switching branches on a test VM. |
13:01 |
csharp_ |
I may variable-ize the Evergreen source dir name to allow for non-Git or alternate repo usage |
13:03 |
csharp_ |
updated the paste with that change |
13:05 |
Dyrcona |
csharp_: Good idea. You might want to add this for bootstrap OPAC: rm -rf $INSTALL_DIR/var/web/opac/deps/node_modules |
13:08 |
Dyrcona |
csharp_: And this one, too: rm -erf $SOURCE_DIR/Evergreen/Open-ILS/web/opac/deps/node_modules |
13:08 |
Dyrcona |
oops. typoed the rm-rf... |
13:09 |
jeffdavis |
not a command you want to typo :) |
13:09 |
berick |
well, "erf!" is the sound you make you typo that one |
13:09 |
csharp_ |
oh right - always forget about that |
13:09 |
csharp_ |
-oof |
13:09 |
csharp_ |
-ohs*t |
13:10 |
berick |
hah |
13:11 |
Dyrcona |
:) |
13:13 |
* Dyrcona |
tries the script out on a test vm, because I was just about to install Lp1947898 for testing with rel_3_10. |
13:13 |
Dyrcona |
Lp 1947898 |
13:13 |
pinesol |
Launchpad bug 1947898 in Evergreen "Enhanced MARC importer script electronic_marc_import.pl" [Wishlist,Confirmed] https://launchpad.net/bugs/1947898 - Assigned to Jason Stephenson (jstephenson) |
13:36 |
csharp_ |
updated the paste with the bootstrap addition |
13:39 |
csharp_ |
@who has already purchased Tears of the Kingdom? |
13:39 |
pinesol |
phasefx has already purchased Tears of the Kingdom. |
13:39 |
csharp_ |
pinesol: I don't doubt it |
13:39 |
pinesol |
csharp_: I'm sorry, Dave. I'm afraid I can't do that. |
13:39 |
csharp_ |
pinesol: not Dave - csharp |
13:39 |
pinesol |
csharp_: Down time is a fact of business when you're a poor 501c3 corporation. |
13:56 |
Dyrcona |
I should modify that dunno to use <who> |
14:12 |
|
jihpringle joined #evergreen |
14:14 |
csharp_ |
Dyrcona: nah - I like the Dave part myself |
14:18 |
|
jihpringle joined #evergreen |
14:27 |
|
kmlussier joined #evergreen |
14:58 |
|
Christineb joined #evergreen |
15:08 |
jeffdavis |
Here's an interesting incident: user successfully logs in to the staff client, and our servers start getting spammed with tens of thousands of open-ils.auth.session.delete requests for their brand-new authtoken. |
15:08 |
jeffdavis |
About 30 seconds later, ejabberd error log shows a message "Killed 1 process(es) consuming more than 9149 message(s) each" - I guess ejabberd decided to kill an open-ils.auth process that was handling the requests? |
15:09 |
berick |
never seen that but I suspect 'process' in this case refers to an ejabberd thread |
15:10 |
jeffdavis |
EG logs then show a message "Removing router class 'open-ils.auth' because of a bad top-level file descriptor", websocketd starts complaining about "WS received XMPP error message" and we start seeing white screens in the client. Fixed by restarting services. |
15:11 |
jeff |
we have observed similar, at least in the spamming session delete requests, though I don't know that we confirmed it was for the new (vs an old) auth token. |
15:14 |
jeff |
one theory was related to background tabs, and/or a laptop resuming/waking from sleep. |
15:14 |
jeff |
it is thankfully (yet annoyingly, also) difficult to reproduce. |
15:18 |
jeffdavis |
We've seen the request spamming before I think, not sure that it failed in this particular way though. I am worried about that log message that says ejabberd killed a process (or thread as Bill says) - I haven't run into that before, and it seems not good for it to be doing that in a way that OpenSRF can't deal with. |
15:18 |
jeffdavis |
(solution: move OpenSRF to Redis :P ) |
15:23 |
jeff |
I think that message is from the ejabberd oom_killer. Of course, setting oom_killer: false in the ejabberd config might result in you actually running out of memory, and the OS OOM killer kicking in... |
15:23 |
jeff |
(which would also potentially introduce a scenario that opensrf can't deal with) |
15:24 |
jeffdavis |
but at least OS-level oom-killer is a familiar problem! |
15:25 |
jeffdavis |
jokes aside though, I'll look into the oom_killer config thing, thanks for mentioning that |
16:39 |
JBoyer |
FYI, with the talk of releases coming out Wednesday and security patches potentially being included I've pushed back 3.11 Beta 1 to that same day. |
16:39 |
Dyrcona |
JBoyer: Sounds good to me. |
16:59 |
pinesol |
News from commits: Translation updates - newpot - main <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=6ac9a0508f31e1a5a6cf20b39974a6f84314b0f3> |
17:06 |
gmcharlt |
Bmagic: noting that I've made some edits to https://wiki.evergreen-ils.org/doku.php?id=poeditor#making_strings_available_for_translation today |
17:16 |
gmcharlt |
Bmagic: also noting that I'm reaching out to dbwells about transferring the translation project to a new TEP organization on POEditor |