| Time |
Nick |
Message |
| 07:24 |
|
collum joined #evergreen |
| 07:30 |
|
kworstell-isl joined #evergreen |
| 08:36 |
|
mmorgan joined #evergreen |
| 08:44 |
|
dguarrac joined #evergreen |
| 09:35 |
|
dguarrac joined #evergreen |
| 11:16 |
|
Dyrcona joined #evergreen |
| 11:18 |
Dyrcona |
Hey, guess what. Login is not working at all for me on the server where I set up MFA today. "TypeError: egCore.auth.provisional is not a function" in addition to a websockets error that might be because I'm not using the VPN... not sure on the latter, but I did check that websocketd is running. |
| 11:27 |
|
smayo joined #evergreen |
| 11:29 |
smayo |
I've got the same problem as Dyrcona |
| 11:31 |
smayo |
And rebuilding my server isn't working because schema.actors.sql can't find some perlmodule and the database rebuild doesn't complete |
| 11:33 |
|
redavis joined #evergreen |
| 11:35 |
|
Christineb joined #evergreen |
| 11:39 |
Dyrcona |
I rebooted the machine, wiped out /openils/var/web/eg2/en-US/ and copied the files back in place. Then, I restarted services, ran autogen.sh, restarted apache2, and verified that Apache2, nginx, websocketd, and OSRF services were running. |
| 11:40 |
Dyrcona |
Opened the Chrome developer tools, cleared my cached, and reloaded the login screen. Other than the header at the top, it's blank. Possibly unhandled rejection talking to websockets, so maybe I'm blocked in a firewall somewhere. |
| 11:40 |
Dyrcona |
I am using my cell phone and having "fun" with the VPN not connecting, either, so..... |
| 11:42 |
Dyrcona |
I did have an issue last week using MFA with the developer tools open and the setting to disable cache while devtools are open enabled. |
| 11:42 |
Dyrcona |
I didn't mention it, 'cause I figured it was one of those things. |
| 11:42 |
redavis |
MFA does rely on setting a cookie, so that might account for the issue. |
| 11:42 |
Dyrcona |
Closed the developer tools and login screen shows up. I'll open them again and reload. |
| 11:43 |
Dyrcona |
cache != cookies, and if it is, devtools are broken. |
| 11:44 |
Dyrcona |
Login screen displays, but I still can't login, and that's probably websockets and the network. |
| 11:44 |
Dyrcona |
I still get this in the console, though: TypeError: egCore.auth.provisional is not a function |
| 11:45 |
redavis |
I know that cache != cookies. But my experience with Chrome is that when it does one thing, it often does another thing that has to be explicitly dealt with. Not saying that's the case in your situation. Just been my experience. |
| 11:45 |
Dyrcona |
I'm gonna disable MFA for my account type in the database to see if it makes a difference. There's also a patch I could try. I think sandbergja reported something like this recently? |
| 11:48 |
Dyrcona |
After doing that, I reload and it asks me to register a workstation just like that. |
| 11:50 |
redavis |
++ |
| 11:52 |
redavis |
It might be worth, if you have time, recreating that errant behavior. It might be a niche bug, but it's going to impact anyone (I think) that have MFA enabled and is also working in a dev environment. |
| 11:52 |
Dyrcona |
I'm going to try the branch from Lp 2080764 to see if that helps. |
| 11:52 |
pinesol |
Launchpad bug 2080764 in Evergreen ""Loading configured secondary authentication factors..." message even when MFA is not on" [Undecided,Confirmed] https://launchpad.net/bugs/2080764 |
| 11:53 |
Dyrcona |
redavis: There's nothing magical about our dev environment. It's a copy of production, and I got the error before enabling devtools this morning, and after verifying that everything was running on the server. I don't think it's a niche problem, necessarily. |
| 11:53 |
redavis |
not sure something can be a regression when it's a new feature. |
| 11:54 |
redavis |
Oh, I mean a dev browser environment...dev tools open, some settings turned on and off to make things show up without rigmarole. That kind of thing. |
| 11:54 |
redavis |
It might not be niche. |
| 11:55 |
Dyrcona |
Heh. Didn't do anything other than set mfa_allowed back to true on my permission group after logging out. I then hit the login button, the "Checking secondary factors" message pops up, and I'm logged in. There's definitely some flakiness going on. |
| 11:57 |
sleary |
I believe it's the conflict between self checkout and MFA that Mike identified in his last comment on Jane's bug. I'm trying out his patch now (rebuilding everything). I'm stuck as well. |
| 11:59 |
|
jihpringle joined #evergreen |
| 11:59 |
redavis |
Ooooooooooh. sleary, that'd be a fun thing to think about if there was no stress involved. |
| 12:00 |
sleary |
smayo I think one of us typo'ed on --primary-disabled-hover |
| 12:02 |
Dyrcona |
I don't have Angular self check on this system, so maybe the issue is resolved by the first commit. |
| 12:04 |
Dyrcona |
That said, it seems to be working without it, even after clearing cache and hard reloading. I should set the interval setting to force it to ask me again. |
| 12:08 |
Dyrcona |
Anyone know how I can force it make me re-authenticate with my second factor? |
| 12:13 |
|
jihpringle39 joined #evergreen |
| 12:14 |
smayo |
sleary either that or one of us got distracted and didn't come back to it |
| 12:15 |
sleary |
seems likely |
| 12:16 |
Bmagic |
eeevil: mmorgan: jihpringle: after adding another copy at another branch, moving the original call number from one branch to another and back, (not changing the item status, it's still in process) - now search is scoping and returing results correctly! WTF? |
| 12:17 |
Bmagic |
that fixed it for that single title, I have a few others that are still messed up |
| 12:17 |
jihpringle39 |
weird |
| 12:18 |
Dyrcona |
Bmagic: Was the call number originally at the wrong place, or did you change to the wrong place and change it back? |
| 12:18 |
Bmagic |
I changed it to the wrong branch and back to the correct branch |
| 12:18 |
mmorgan |
Bmagic: Did you ever check the deleted flag on the original call number record for the item? |
| 12:19 |
Dyrcona |
On MFA: After all my fooling around this morning, it's working without any extra patches. Maybe this server is just flaky? |
| 12:19 |
Bmagic |
deleted flag has never been set to true for the call number, item, shelving location, and opac_visible is true also all the way down |
| 12:20 |
Dyrcona |
Updating location causes some triggers to fire. I'm guessing whatever data was maintained by those triggers was stale. |
| 12:21 |
Bmagic |
what led me here was testing a different concerto test server, I added an item (in processing) and the search results are coming up (on the test machine) the way I would expect. Which rules out the item status as being the culprit. An item can be "In Process" and still show up in search results when scoped at the branch where it's cataloged |
| 12:21 |
Dyrcona |
I made MFA required for my patron group. (Aside: It's fun when the 6 digit code changes while you're entering it on the site.) |
| 12:24 |
eeevil |
Dyrcona: re patron mfa, it def should not "work" for opac logins ... unless you customize your opac code to send "staff" as the login type! :) |
| 12:28 |
Dyrcona |
eeevil: I'm fiddling with the staff client only. |
| 12:31 |
Dyrcona |
eeevil: Is there a way that I can make it prompt me every time I login when mfa is allowed, but not required for my group? It didn't prompt me even though I had it set up. |
| 12:32 |
Dyrcona |
Also bad time to ask a question, when I'm thinking about walking away to get some lunch. |
| 12:33 |
Dyrcona |
Maybe I'll set it back to just allowed, sign out, get some lunch, and see what happens when I get back. I set the expiration interval to 1 minute. |
| 12:34 |
mmorgan |
Maybe ' |
| 12:34 |
mmorgan |
Get Lunch' should be an additional option for a factor :) |
| 12:34 |
Dyrcona |
:) |
| 12:35 |
Dyrcona |
I'll be signing out while I get lunch. |
| 12:35 |
Dyrcona |
of chat, I mean. I usually don't but today I will. |
| 12:35 |
eeevil |
allowed is "can configure, and prompted to do so until that's done" ... it's for a soft rollout. required is "you must use it (unless you already did within the interval), and must configure it if you haven't" |
| 12:36 |
Dyrcona |
eeevil: If you've configured it, is there anyway for it to ask you again as a user? |
| 12:36 |
Dyrcona |
I mean if I turn it on, I expect to be asked to use it. |
| 12:36 |
eeevil |
Dyrcona: make it required ... |
| 12:36 |
Dyrcona |
Ah, but what if I want it to be optional so staff can choose? I guess that's not in the plans? |
| 12:37 |
eeevil |
not for v1 |
| 12:37 |
Dyrcona |
I know, for staff that doesn't make much sense, really. For patrons, though... |
| 12:37 |
Dyrcona |
OK for later, then! eeevil++ |
| 12:37 |
* Dyrcona |
moseys off to find something to eat. |
| 12:38 |
Bmagic |
new info: I updated the call number directly in the database this time. update asset.call_number set owning_lib=102 where id=xxxx. Then checked search results. the results came back for the branch. Then I updated the call number in the database again, back to the original owning_lib, and boom! search is returning the bib now. OMG |
| 12:38 |
redavis |
Bmagic++ |
| 12:39 |
Bmagic |
not really |
| 12:39 |
Bmagic |
redavis: the question is bhy! |
| 12:39 |
Bmagic |
why! |
| 12:39 |
Bmagic |
I haven't figured out why |
| 12:39 |
Bmagic |
since it fixed search after updating the call_number row directly, then it has* to be a DB trigger right? |
| 12:40 |
redavis |
I have no expertise nor qualifications to help you answer that question. Although it could be that you somehow circumnavigated a stale data issue like Dyrcona described. |
| 12:40 |
redavis |
And, I appreciate you believing that I could possibly understand DB triggers at anything other than a conceptual level...but I don't. |
| 12:40 |
Bmagic |
TRIGGER z_opac_vis_mat_view_tgr |
| 12:41 |
mmorgan |
Bmagic: Sounds likely, Dyrcona referenced db triggers earlier |
| 12:42 |
Bmagic |
so, maybe to "fix" it, I need to execute: update asset.call_number set id=id where owning_lib=103 and not deleted; |
| 12:42 |
Bmagic |
that should fire the asset.cache_copy_visibility() funcation for all* of the call numbers that are hooked up to this branch that never seems to return search results |
| 12:43 |
Bmagic |
nope, I don't think that will work, after reading the function |
| 12:44 |
eeevil |
Bmagic: you should look into why that trigger was disabled or skipped. I assume this is an incoming migration? did you (or some tool you use) disable triggers for speed? a trigger doesn't just not rung... |
| 12:44 |
Bmagic |
it will only execute UPDATE asset.copy_vis_attr_cache when OLD.record <> NEW.record OR OLD.owning_lib <> NEW.owning_lib |
| 12:44 |
eeevil |
run, even |
| 12:44 |
Bmagic |
hmmmmm, yes, migration |
| 12:45 |
redavis |
Ohh, migration. #fightthegoodfight |
| 12:45 |
Bmagic |
I think this is it, the trigger was disabled for speed! I bet |
| 12:45 |
Bmagic |
foggy memories |
| 12:45 |
Bmagic |
coming into focus.... |
| 12:47 |
eeevil |
just to wrap it up for the logs, then ... nothing to do with search, rather, (likely) tricksy migration data loading procedure caused normal EG stuff to be skipped |
| 12:48 |
Bmagic |
ok, let's assume that's the deal. Can I affect the DB to re-visible-cache calculate a set of call numbers? Or I don't mind doing the whole db |
| 12:58 |
mmorgan |
Bmagic: Just browsing functions, asset.calculate_copy_visibility_attribute_set(bigint) takes a copy id. Haven't read through it, though. |
| 13:05 |
Bmagic |
I was looking at that function - can't use it, because it doesn't update the database, It's a function that other functions use to get the INT[] array that eventually makes it's way into asset.copy_vis_attr_cache |
| 13:07 |
Bmagic |
maybe I can create my own function temporarily that executes the portion of asset.cache_copy_visibility() that I want |
| 13:39 |
eeevil |
Bmagic: why can't you just do something like: update asset.copy_vis_attr_cache set vis_attr_vector=asset.calculate_copy_visibility_attribute_set(target_copy) where [some where-clause-here]; |
| 13:39 |
Bmagic |
thank you, yes, that's what I was trying to think of |
| 13:40 |
Bmagic |
I was concerned that a visibility row may not even exist yet, but I can't find a single call number that isn't represented in the visibility table. so I think I'm looking at only updates |
| 13:40 |
Bmagic |
I found a query that seems to reveal the problem records, and wow, there's only 10 |
| 13:42 |
|
smayo joined #evergreen |
| 13:50 |
Bmagic |
eeevil++ |
| 14:10 |
pinesol |
News from commits: quick syntax error fix for seed file <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=6945458b5a0d93e60b1ef034b7cf1122ffe26ff5> |
| 14:12 |
Bmagic |
mmorgan++ Dyrcona++ redavis++ jihpringle++ # meesa love ma communita |
| 14:17 |
redavis |
Aww, likewise Bmagic++ |
| 14:26 |
|
Dyrcona joined #evergreen |
| 14:38 |
|
kworstell-isl joined #evergreen |
| 14:58 |
Dyrcona |
Nope. Didn't do anything but log out, close a browser window, open a new window and attempt to login, and no bueno. |
| 15:01 |
Dyrcona |
This may be where I give up and say this doesn't work on 3.13. :) |
| 15:02 |
Dyrcona |
There are conflicts over stuff that gets deleted..... It must be subtly different in 3.13. |
| 15:02 |
redavis |
Which not working? MFA? |
| 15:03 |
Dyrcona |
Yeah, MFA breaks login again with the type error message. |
| 15:03 |
Dyrcona |
Seems like closing the tab/window is enough to do it. |
| 15:06 |
Dyrcona |
I'm trying the first patch from Lp 2080764, and it did not apply cleanly, but it looks obvious how to resolve it. |
| 15:06 |
pinesol |
Launchpad bug 2080764 in Evergreen ""Loading configured secondary authentication factors..." message even when MFA is not on" [Undecided,Confirmed] https://launchpad.net/bugs/2080764 |
| 15:06 |
|
kworstell-isl joined #evergreen |
| 15:08 |
Dyrcona |
Guess not. It fails to compile with ng build: Property 'authDomain' does not exist on type 'AuthService'. |
| 15:09 |
Dyrcona |
Ah! Maybe I grabbed the wrong patch. |
| 15:11 |
Dyrcona |
Nope. Correct one. It won't work unless I backport Angular self check I guess. I'll revert the commit and resign myself to turning MFA off on this machine. |
| 15:21 |
Dyrcona |
I could probably mangle it into something that will not throw errors, but then I'm not sure it will resolve my issue. I've disabled MFA and chalked it up to "It doesn't work with 3.13 as mentioned previously." |
| 15:22 |
redavis |
I guess I didn't realize you were trying to backport it into 3.13. |
| 15:23 |
sleary |
Dyrcona I got that message until I restarted services |
| 15:23 |
sleary |
but I'm on main, haven't tried 3.13 |
| 15:29 |
Dyrcona |
sleary: I restarted services earlier today. Looks like setting mfa to false for the group resolves it for me without restarting services. I ran autogen.sh, etc., earlier. |
| 15:29 |
Dyrcona |
It worked for a while. Then I closed the browser window after logging out. Opened a new window and tried to log in and login failed with that console message. |
| 15:30 |
Bmagic |
mmorgan: are you building today? |
| 15:30 |
mmorgan |
Bmagic: Yes. Working on it now. |
| 15:31 |
Bmagic |
ok! I was wondering, of course, for myself |
| 15:31 |
Dyrcona |
mmorgan++ Bmagic++ |
| 15:31 |
Bmagic |
sounds like I'll get crackin on 3.13.4 then! |
| 15:31 |
Dyrcona |
abneiman++ |
| 15:31 |
mmorgan |
:) |
| 15:32 |
* mmorgan |
is still plodding/puzzling through the process. You'll probably have yours built before me! |
| 15:33 |
Bmagic |
you're on |
| 15:44 |
redavis |
Bmagic++ mmorgan++ abneiman++ jsandberg++ holler when y'all are ready and weblike updates shall commence. |
| 15:48 |
mmorgan |
Bmagic: (Or anyone else!) After working through the translations, Part 2: https://wiki.evergreen-ils.org/doku.php?id=dev:release_process:evergreen:how_to_build#translations_part_2update_pofiles |
| 15:48 |
mmorgan |
I have no po file changes to commit. Is that plausible? |
| 15:48 |
Dyrcona |
It is plausible. |
| 15:49 |
Bmagic |
did you run the python script after make newpot? |
| 15:49 |
mmorgan |
Bmagic: Yes. |
| 15:49 |
Bmagic |
and that's where it had no changes? |
| 15:49 |
|
jihpringle joined #evergreen |
| 15:50 |
Dyrcona |
That's the purpose of the script: to get rid of unnecessary changes caused by newpot. |
| 15:50 |
Dyrcona |
more or less. |
| 15:51 |
Dyrcona |
It used to be a manual process and was it ever slow. |
| 15:51 |
Bmagic |
right, for me, it reduced 9 files to 5 |
| 15:51 |
Bmagic |
(on 3.13) |
| 15:55 |
mmorgan |
Bmagic: it was after running this that there were no changes: Evergreen/build/i18n/scripts/update_pofiles -l translation-export-3.12 -e Evergreen |
| 15:56 |
Bmagic |
did you happen to have already cloned the LP bzr repo? |
| 15:56 |
Bmagic |
if so, then, perform a bzr pull to make sure your copy is synced |
| 15:57 |
mmorgan |
Bmagic: I cloned it today for the first time. It failed a few times, but finally cloned successfully. Then I did a bzr pull to make sure. |
| 15:57 |
Bmagic |
perfect |
| 15:58 |
Bmagic |
so, I would say, then that 3.12 hasn't received any patches that required new translations from LP, and in turn didn't prompt our translators to provide the translations |
| 16:00 |
mmorgan |
Ok, thanks! I will proceed then! |
| 16:00 |
mmorgan |
Bmagic++ Dyrcona++ |
| 16:00 |
Bmagic |
mmorgan++ |
| 16:01 |
Bmagic |
and on the 18th day (of the month) he said: "Let there be point releases!" |
| 16:02 |
mmorgan |
s/he/they ;-) |
| 16:04 |
Bmagic |
lol, thanks, yes! I prefer my deity to be female :), gotta deprogram my brain from using "traditional" language |
| 16:08 |
|
kworstell-isl joined #evergreen |
| 16:14 |
Bmagic |
if anyone wants to test the 3.13.4 tarball: https://evergreen-ils.org/downloads/Evergreen-ILS-3.13.4.tar.gz |
| 16:21 |
Dyrcona |
Oh! That reminded me of something I wanted to mention yesterday. If you're looking for any of my pastes on Pastebin.com, they're gone. I deleted my account this weekend because they have been handling their spam issue very badly and deleted several of my innocuous pastes. |
| 16:21 |
Bmagic |
I've started introducing expiration of like 1 day each time I use that site |
| 16:21 |
Dyrcona |
They don't respond when I use the form to say that the paste was removed in error and should be reinstated. |
| 16:21 |
Bmagic |
but, I also don't have an account |
| 16:22 |
Dyrcona |
I had a paid account. It was one of the few sites I've actually paid for. |
| 16:23 |
Dyrcona |
I think the "they" is plural in this case since there are multiple people building point releases. :) |
| 16:23 |
Dyrcona |
I'm not sure I'll have the time to test a tarball today. My circumstances are a bit unusual. |
| 16:43 |
|
kworstell-isl joined #evergreen |
| 18:05 |
|
jihpringle49 joined #evergreen |
| 18:12 |
|
mmorgan left #evergreen |
| 19:38 |
|
Dyrcona joined #evergreen |
| 19:40 |
Dyrcona |
So, looks like sandbergja is going to test the 3.13.4 tarball. I've got an hour right now and thought I would give it a go. I have to set up VMs on my laptop again after reinstalling the O/S. |
| 19:49 |
Dyrcona |
Downloading the latest ISOs for Debian 11 and 12, and Ubuntu 22.04 and 24.04 looks like it will take up the rest of the hour. |
