IRC log for #koha, 2017-07-20

All times shown according to UTC.

Time S Nick Message
00:00 jamesb joined #koha
00:13 Francesca joined #koha
00:19 Francesca joined #koha
00:41 Scott-CSPL joined #koha
00:46 deb-CSPL joined #koha
00:54 caboose-afk joined #koha
02:06 Francesca joined #koha
02:59 Francesca joined #koha
03:02 Francesca joined #koha
03:03 kathryn joined #koha
05:23 cait joined #koha
06:10 laurence joined #koha
06:21 reiveune joined #koha
06:22 reiveune hello
06:29 LibraryClaire joined #koha
06:29 fridolin joined #koha
06:29 fridolin hie there
06:37 marcelr joined #koha
06:37 marcelr hi #koha
06:46 AndrewIsh joined #koha
07:30 gaetan_B joined #koha
07:30 gaetan_B hello
07:30 paul_p joined #koha
07:41 Francesca joined #koha
07:52 gandy joined #koha
08:01 kathryn joined #koha
08:46 cait joined #koha
09:23 Francesca joined #koha
10:22 kidclamp_away joined #koha
10:28 Francesca joined #koha
10:30 Bhootrk_ joined #koha
10:32 Bhootrk_ joined #koha
10:33 Bhootrk_ joined #koha
10:35 Bhootrk_ joined #koha
10:49 * cait waves
10:51 marcelr o/
10:51 marcelr cait++ # old_issues
10:52 cait marcelr: not sure i deserve that... it feels like whatever we do, things are not going to be unproblematic
10:52 cait but safest bet for me now seems, stop the renumbering, use the mysql fix
10:52 cait and maybe provide a script to clean up data
10:53 marcelr yeah, its ugly stuff
10:53 cait renumber and set PKs once correctly and then leave it to mysql again
10:53 cait i'd appreciate someone looking at my finding and commenting
10:53 cait i will probably push the release a few days to get this sorted
10:53 marcelr Joubu will certainly do :)
10:54 cait yeah, but I think we put our brain power into it, fresh input would be good
10:54 cait it's not a 2 person problem ;)
11:29 meliss joined #koha
11:45 caboose joined #koha
12:08 papa joined #koha
12:09 edveal joined #koha
12:12 nengard joined #koha
12:24 oleonard joined #koha
12:25 rsantellan joined #koha
12:25 LeeJ joined #koha
12:25 rsantellan good morning #koha
12:25 * LeeJ waves
12:25 LeeJ hi #koha
12:25 * cait waves
12:28 LeeJ cait: hiiiiiiiiiiii
12:33 jzairo joined #koha
12:36 cait :)
12:36 * cait waves at jzairo too
12:36 jzairo morning :)
12:40 LeeJ morning jzairo
12:41 jzairo hi LeeJ!
12:54 LeeJ cait: going to try to test out kohadocs in terminal to see how far I can push it..mostly to see if I can automate the process of the gitlab ssh key
12:57 Joubu hi #koha
12:58 LeeJ hi Joubu
13:02 Joubu cait, fridolin: I would suggest you to postpone the next release until we have a solution for bug 18651, and ask for help on koha-devel
13:02 huginn Bug[…]_bug.cgi?id=18651 blocker, P5 - low, ---, jonathan.druart, Pushed to Stable , Move of checkouts is still not correctly handled
13:03 Joubu I did all what I could
13:03 cait Joubu++ i know, sorry
13:04 cait LeeJ: sounds good - i have seen there are also tools to open up merge requests from command line
13:04 cait that might be interesting along the road too
13:04 cait not native gitlab feature, but there are some external programs which seem able to do that, didn't get to try it out
13:04 LeeJ cait: do you have a link or remember where you found those?
13:04 cait yeah... but on the other laptop :( let me do a quick search
13:05 LeeJ cait: it's okay...I can search! Just thought I'd check :)
13:05 cait[…]
13:05 cait there was one tht seemed kinda easy
13:05 cait looking if i can spot that quickly
13:05 cait this is the api doc i think
13:06 cait and I think i was looking at this one to test:
13:09 kellym joined #koha
13:09 Joubu cait: I am not going to send an email to koha-devel to ask for help. I am asking for help/test/feedback for 2 months now and it did not get the expected attention
13:09 cait Joubu: i will
13:09 cait there has been opened another bug today too
13:10 cait 18963
13:10 cait I was not able yet to figure out how it happens
13:10 Joubu It's intolerable to have a *blocker* that affect stable releases waiting for 2 months
13:10 fridolin Joubu: we may revert on stable branches ?
13:10 cait it's similar to the other email the other week
13:10 cait bug 18963
13:10 huginn Bug[…]_bug.cgi?id=18963 enhancement, P5 - low, ---, koha-bugs, NEW , Check In Issue
13:11 cait i think the problem is picking the best solution as there is now way to have a complete fix right now
13:11 cait atm i'd favor pushing for the mysqlfix- I have heard rangi tested it? hope to test it too, but fridolin, maybe you could take a look as well?
13:11 fridolin Joubu: ah heu no I'm wrong, its about the issues > old_issues
13:11 fridolin indeed release can wait
13:11 cait Joubu: what what the lock trying to fix?
13:12 cait i think renumbering was implemented with 18242, but without accountlines? is that right?
13:12 cait i am still trying to understand the code better
13:12 JoshB joined #koha
13:12 tcohen_ joined #koha
13:12 cait tcohen_: another volunteer!
13:12 fridolin i did not look very close, not sure i can take the time
13:12 fridolin but i'll try
13:13 tcohen_ morning
13:13 cait as Joubu said
13:13 cait it's a circulation blocker
13:13 Joubu fridolin: when you will have 50 customers crying because they cannot checkin, I am sure you will find time
13:13 cait look at bug 18963 and the email thread i linked... this is seriously damaging us
13:14 cait we need to make time for this
13:14 Joubu cait: basically, the safe way is:
13:14 Joubu 1. Start a transaction
13:14 fridolin Bug 18242 fixes a case that is realy not usual i'd say
13:14 huginn Bug[…]_bug.cgi?id=18242 critical, P5 - low, ---, jonathan.druart, Pushed to Stable , Move of checkouts to old_issues is not handled correctly
13:14 tcohen1 joined #koha
13:14 Joubu 2. LOCK the table # To make sure nobody will read id and avoid race conditions
13:15 JoshB joined #koha
13:15 francharb joined #koha
13:15 Joubu 3. Move the content from one table to the other, dealing with ids
13:15 Joubu 4. UNLOCK the table
13:15 Joubu 5. Commit # end of transaction
13:15 Joubu this is the only way (I know) to avoid race condition and data lost
13:16 tcohen pfiu
13:16 Joubu the problem is that the lock/unlock cause problem with dbix: you have to name the table you are going to LOCK
13:16 Joubu but DBIC uses aliases (me) for tables, and MySQL does not like that
13:17 Joubu see 905572910b3af03d86fdbd3e52c72614f41c978b in master
13:17 marcelr poor dbix bug?
13:17 talljoy joined #koha
13:18 cait the problem is ee is that that the check-in is rejected
13:18 cait I assume, while the table is locked
13:18 cait so the locking seems problematic - especially since we were trying so hard to speed things up in circ
13:19 cait as I wrote, I am also worried what would happen at the self check
13:19 tcohen we shouldn't lock, because it, well, locks inserts
13:19 cait batch checkin is pretty fast...
13:19 cait yeah
13:19 tcohen MySQL doesn't have row level locking
13:19 Joubu People (all of you) HAVE TO test and confirm the fix at DBMS startup fixes the issue
13:19 cait yep
13:20 tcohen true
13:20 cait and then we have to propagate that
13:20 Joubu in the next few days
13:20 cait and could possible build back the code and remove 18424 - it hink that's the problem maker now for the bug reports we see
13:20 cait but so far... suspicion
13:20 Joubu then I can provide a fix (2 actually, 1 for master/17.05 and 16.11) to handle the error correctly
13:20 Joubu but if it is not done quickly, somebody else will have to do it
13:21 cait can we outline how we expect the fix to be?
13:22 cait i made some tests last night reverting 18242 - looked ok in the first tests
13:22 marcelr the idea to get back at the old ids in sync does not sound too bad; and if a id would collide, make a choice (pick new one)?
13:22 cait i think we either revert htat or need to make the 'renumbering process' set the PK correctly
13:22 cait otherwise the problem moves on to all following issues
13:22 kellym joined #koha
13:23 marcelr the renumbering stuff did not work out
13:23 cait atm i am quite suspicious of it, but i can't prove it causing the problems in 18963 and on the mailing list
13:23 Joubu cait: it's ok reverting 18242?
13:23 Joubu what do you mean?
13:23 cait it worked cleanly at least
13:24 Joubu then I did all this stuff for nothing
13:24 cait but what happens is that if you have a dobule up, the old-issues move fails silently
13:24 cait the good side is... that after that it's over. with the renumbering, the PK are never fixed
13:24 marcelr that problem is not as big as what we have now?
13:24 cait issues keeps producing numbers that are alredy in old_issues
13:25 cait if I have no big thinking error there
13:25 cait yeah, that was my thinking marcelr
13:25 tallerjoy joined #koha
13:25 cait it might be tolerable
13:25 cait because the real fix is merging the tables
13:25 cait and that's nothing we can do for stable releases
13:26 Joubu cait: without 18242 you should notice data lost
13:26 cait i know
13:26 cait i just wonder if it's the lesser evil
13:26 cait given we will also publish the mysql based fix
13:27 cait and maybe we could throw a script together that renumbers existing duplicate issue_ids and sets the PKs correctly
13:27 marcelr we could run this in cleanup_database too
13:27 cait clean up and prevention
13:27 cait at the source
13:27 marcelr each day or week
13:27 Joubu to me it's the worst, you cannot recover it
13:28 Joubu see bug 18931
13:28 huginn Bug[…]_bug.cgi?id=18931 enhancement, P5 - low, ---, jonathan.druart, Needs Signoff , Add a "data corrupted" section on the about page
13:28 tallerjoy joined #koha
13:28 Joubu so, what next?
13:29 marcelr write a patch and i qa it tomorrow :)
13:29 Joubu nope
13:29 cait yeah but it also happens for reserves
13:29 cait items
13:29 cait everything else
13:29 wahanui it has been said that everything else is just extras
13:30 cait i feel like we tried to repair it, but it's a dead end (again because of mysql not supporting the kind of locking we'd need)
13:30 marcelr joubu: file locking ?
13:30 Joubu my plan was: People (all of you) HAVE TO test and confirm the fix at DBMS startup fixes the issue, in the next few days. Then I can provide a fix (2 actually, 1 for master/17.05 and 16.11) to handle the error correctly. But if it is not done quickly, somebody else will have to do it
13:30 cait Joubu has said he can't write it right now
13:31 cait i can push the release another week...
13:31 cait but i would not like to wait much longer, just because we really need to fix this
13:31 cait pushing the release shoudl not be an endless excuse, we need to motor
13:37 caboose-biab joined #koha
13:38 tallerjoy joined #koha
13:39 kholt joined #koha
13:40 francharb joined #koha
13:45 caboose joined #koha
13:47 cait seriously, is someone going to test the mysql based fix now?
13:49 kchris joined #koha
13:50 oleonard well I'm failing at fixing all the broken things I'm trying to fix today, so maybe I can successfully test something.
13:50 oleonard Sorry I haven't been paying attention. What's the bug number?
13:50 marcelr cait: although the fix looks kind of silly, it should probably work
13:51 marcelr oleonard: it is the dbms fix on the wiki somewhere
13:51 oleonard I see it.
13:51 oleonard https://wiki.koha-community.or[…]uto_increment_fix
13:51 marcelr yes
13:51 cait oleonard: check out books, check them in - restart mysql
13:52 cait check out a book again, notice that a low issue_id is generated in issues (that already exists in old_issues)
13:52 cait appply the fix, do some testing again, it should not happen reset the PK again in issues
13:52 kmlussier joined #koha
13:52 cait and try for other tables too i think, like reserves
13:53 cait Joubu: what's with items, deleteditems etc?
13:53 cait shoudl we also add that to the mysql fix?
13:54 Joubu If I forgot it, yes
13:55 Joubu I forgot it indeed
13:58 jac joined #koha
14:20 caboose joined #koha
14:24 oleonard "Localize the [mysqld] section"  Does that mean "Locate?"
14:26 Joubu means "search for the [mysqld] section"
14:26 Joubu /mysqld
14:30 caboose joined #koha
14:33 jbeno joined #koha
14:43 fridolin see u
14:43 fridolin left #koha
14:55 LibraryClaire left #koha
15:00 TGoat joined #koha
15:03 oleonard Joubu: In testing the mysql fix I have still gotten a re-used biblionumber in the process
15:05 cait a newly catalogued one after mysql restart?
15:07 oleonard Yes. Is there a way to confirm that the fix has been correctly applied? Perhaps I messed it up.
15:08 cait we are trying now too, maybe can say more in a bit
15:18 NateC joined #koha
15:48 caboose joined #koha
15:52 caboose joined #koha
15:56 rocio joined #koha
16:06 caboose joined #koha
16:13 caboose joined #koha
16:17 caboose joined #koha
16:18 kellym joined #koha
16:22 caboose joined #koha
16:26 cait left #koha
16:36 laurence left #koha
17:05 caboose joined #koha
17:18 Joubu cait: bug 18966
17:18 huginn Bug[…]_bug.cgi?id=18966 blocker, P5 - low, ---, jonathan.druart, ASSIGNED , Move of checkouts - Deal with duplicate IDs at DBMS level
17:18 Joubu @later tell cait bug 18966
17:18 huginn Joubu: The operation succeeded.
17:18 Joubu and all devs ^
17:30 thd joined #koha
17:30 thd pianohacker: are you present?
17:31 rocio_ joined #koha
17:32 tallerjoy joined #koha
17:42 tallerjoy joined #koha
18:29 francharb joined #koha
18:46 francharb joined #koha
18:56 caboose joined #koha
19:03 oleonard joined #koha
19:19 Scott-CSPL joined #koha
19:19 deb-CSPL joined #koha
19:59 caboose joined #koha
20:05 cait joined #koha
20:13 rangi @later tell Joubu the mysql fix works for me, the mysql fix and removing 18242 gets back a usable system
20:13 huginn rangi: The operation succeeded.
20:14 Joubu rangi: thanks! Seen bug 18966?
20:14 huginn Bug[…]_bug.cgi?id=18966 blocker, P5 - low, ---, jonathan.druart, Needs Signoff , Move of checkouts - Deal with duplicate IDs at DBMS level
20:15 Joubu the idea is to revert all the sh**, keep the transaction to avoid data lost, then display a warning and link to the wiki page if the dup ID happens
20:16 rangi reading now
20:17 rangi yes, thats exactly the way I would do it too
20:18 rangi ill try to get a signoff on it
20:18 rangi but im the only one in the office today, so might not be able to
20:18 cait what happens with the issue, did we figure that out yet?
20:18 cait just trying to catch up
20:18 rangi i care way way way less about fixing old stuff
20:18 rangi than stopping it getting worse
20:18 rangi id rather stop it getting worse (most people in the world have still not upgraded)
20:19 rangi and then worry about fixing duplicates
20:19 cait rangi: i meant if a duplicate already is there - we have to do something to get it checked-in
20:20 rangi yes
20:20 rangi but lets worry about that on a different bug
20:20 cait and of course, prevent it from happenign is the best way
20:20 rangi and stop it happening first
20:20 cait but we already have some libraries asking on how to deal with it once it's happened
20:20 rangi yes, and I dont care
20:20 rangi :)
20:20 cait hm
20:20 rangi as much as I care about stopping it for 10k other libraries
20:21 rangi it can be a different bug
20:21 cait heh
20:21 rangi fixing duplicate ids
20:21 rangi lets not make this fix take even longer
20:21 cait i think we want the same thing, suspect we are misunderstanding each other
20:21 cait will go back to read the bugs
20:21 rangi but making it do something like try and fix old duplicates
20:21 rangi which is error pronbe
20:21 cait nah
20:21 cait that's not what i meant
20:21 rangi lets just stop it making them
20:21 cait just you need to be possible to get it off the patron account
20:21 rangi yes
20:21 cait even if we put some sql somewhere or so
20:21 rangi different bug
20:21 rangi not this one
20:22 cait k
20:22 rangi dont hold this one up, because of that
20:22 cait was more thinking of putting it on the wiki page for now
20:22 cait that can happen independently
20:22 rangi thats fine, but dont let it take time away from fixing this ;)
20:22 rangi perfect is the enemy of good
20:22 cait heh
20:23 cait i have spent a good time of this day trying to find out a way of this mess...
20:23 cait Joubu++
20:23 rangi Joubu++
20:24 rangi did you see it even has a data problems bit?
20:24 Joubu now that we have a proper fix, so we will not need to do it at code level
20:24 rangi yeah
20:25 Joubu I think summarize what has been done (badly), what have to be done (now) and what will be done, later
20:25 rangi sounds good to me
20:25 Joubu the TODO LATER part will be to fix the broken DBs
20:25 rangi *nod*
20:25 Joubu mysql--
20:27 cait yeah
20:27 cait can postgres do this right? or even mariadb?
20:27 cait doesn't help us... just wondering
20:27 Joubu mariadb behaves like mysql, no idea for postgres
20:27 rangi mariadb could, but we run it like it was mysql
20:28 rangi its to do with innodb tables
20:28 rangi not the db itself
20:28 rangi if that makes sense
20:28 rangi if those tables were still old myisam
20:28 rangi it wouldnt have happened either
20:29 cait ah
20:29 cait thx
20:29 rangi (we'd have other problems instead :))
20:30 cait yeah, just trying to learn... it's horrible, so it least try to learn something heh
20:30 rangi but yeah it is the innodb behaviour to forget its increment
20:31 rangi i dunno if xtradb does (innodb in mariadb does this too)
20:31 rangi but yeah that was going to be my last suggestion, if the config fix didnt work
20:31 rangi we change the table type
20:34 Joubu rangi: I think the wording on the about page will need to be adjusted ;)
20:34 rangi :)
20:47 cait will catch up with the patches in a bit - have to pack first, but keep going! :)
20:51 rocio_ joined #koha
20:59 reiveune bye
21:00 reiveune left #koha
21:01 bag bug 18966
21:01 huginn Bug[…]_bug.cgi?id=18966 blocker, P5 - low, ---, jonathan.druart, Needs Signoff , Move of checkouts - Deal with duplicate IDs at DBMS level
21:05 Freddy_Enrique joined #koha
21:05 * Freddy_Enrique vows to everyone
21:06 * cait waves
21:07 Freddy_Enrique Nice to mmet you all :). New in this forum
21:07 caboose joined #koha
21:08 Freddy_Enrique Well, i ve got one question... I think I registered to the general email list
21:08 Freddy_Enrique I sent one question a couple of weeks ago, and nothing happened
21:09 Freddy_Enrique any insights?
21:11 Joubu Hi Freddy_Enrique
21:11 Joubu what was the question?
21:12 Freddy_Enrique Up until now I've been installing koha using my VM ip
21:12 Freddy_Enrique I dont really understand the second method
21:12 Freddy_Enrique I heard that the IP should be used for test purposes, but the other one for production
21:14 Joubu Freddy_Enrique: I have to leave, I do not see the question on the ML, I see one topic "OPAC CUSTOMIZATION" from May
21:14 Joubu please resent if you do not get an answer here
21:15 Freddy_Enrique yeah!!! that was my question
21:16 Freddy_Enrique sure
21:20 kathryn joined #koha
21:24 kholt left #koha
21:49 jzairo joined #koha
21:50 cait joined #koha
22:30 JoshB left #koha
22:34 caboose joined #koha
22:38 caboose joined #koha
22:58 cait1 joined #koha
23:04 caboose joined #koha
23:35 papa joined #koha
23:37 dilan joined #koha
23:47 dilan joined #koha

| Channels | #koha index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary