Evergreen ILS Website

IRC log for #evergreen, 2015-07-13

| Channels | #evergreen index | Today | | Search | Google Search | Plain-Text | summary | Join Webchat

All times shown according to the server's local time.

Time Nick Message
02:10 TaraC joined #evergreen
04:28 _robbat2|irssi joined #evergreen
04:51 pinesol_green Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html>
05:34 Mark__T joined #evergreen
07:44 rjackson_isl joined #evergreen
07:48 jboyer-isl joined #evergreen
08:11 mrpeters joined #evergreen
08:27 ericar joined #evergreen
08:33 collum joined #evergreen
08:35 mmorgan joined #evergreen
08:44 akilsdonk joined #evergreen
08:59 Shae joined #evergreen
09:24 kmlussier dbwells++ # bug 1064679
09:24 pinesol_green Launchpad bug 1064679 in Evergreen "Hourly due dates and fines ignore library closings" (affected: 4, heat: 26) [Medium,Confirmed] https://launchpad.net/bugs/1064679 - Assigned to Dan Wells (dbw2)
09:25 mmorgan dbwells++ indeed!
09:31 maryj joined #evergreen
09:38 yboston joined #evergreen
09:47 pmurray_away joined #evergreen
09:47 mmorgan miker: dbs: or anyone else who might know :) - I've been looking at Opensearch - is it able to return holdings? Do I just need to pick the right format?
09:49 mmorgan oops! NM just answered my own question :-[ -- Yes.
09:50 jboyer-isl mmorgan: no need for :-[ , now you've answered the question for Google users later. ;)
09:51 mmorgan So I did! :)
09:53 pmurray joined #evergreen
10:01 mnsri joined #evergreen
10:03 jihpringle joined #evergreen
10:05 ericar_ joined #evergreen
10:22 Bmagic Would it be typical for EDI orders to go out under one edi account and the invoices to match the same purchase order using a different edi account? (these are different vendcode accounts for the same provider)
10:29 csharp Bmagic: maybe someone linked an invoice to a PO with the wrong vendor/provider?
10:29 * csharp noticed room for that when troubleshooting recently
10:29 Bmagic csharp: The invoice is manually matched? These are EDI invoices
10:30 csharp I know there's a way to link an invoice to an order
10:30 b_bonner joined #evergreen
10:30 rashma_away joined #evergreen
10:30 mnsri_away joined #evergreen
10:35 csharp Bmagic: screenshot: http://picpaste.com/linking_invoices-S7tKlJU5.png
10:36 csharp you have to know the invoice number and the provider - the system doesn't seem to care whether the provider is actually the one to send the EDI invoice
10:37 Bmagic csharp: that step is filled in by EDI
10:38 collum joined #evergreen
10:44 csharp hmm - could someone on 2.7(-ish) please verify whether expired staff accounts can log in to the staff client?
10:53 bshum csharp: I doubt that they can.
10:53 bshum csharp: Couple years ago, I wrote a query to update all our staff accounts to expire several decades into the future to avoid situations like that.
10:54 bshum Once we realized that some of our earlier libraries started having failed logins.
10:54 csharp well, they can for us, so I'm trying to rule out the possibility that this is an EG bug or something in our setup
10:54 * bshum can double check
10:54 csharp bshum++ #thanks!
10:56 bshum csharp: When I use an expired staff account, I can still log into the staff client with it.
10:56 bshum That's fun.
10:56 bshum At least on one of my master test servers.
10:56 csharp excellent - it's not just me!
10:57 bshum So I presume that qualifies your 2.7+
10:57 csharp @bugforeveryoneorjustme.com
10:57 pinesol_green csharp: Evergreen Command Center http://apod.nasa.gov/apod/image/1204​/EndeavourFlightDeck_cooper_1050.jpg
10:57 bshum @bugorfeature
10:57 pinesol_green bshum: It's a bug!
10:57 bshum Hehe
11:03 RoganH joined #evergreen
11:19 jeff yes, expired accounts (of all kinds) can log in. they just can't check items out.
11:20 jeff open-ils.auth has no concept of "expired"
11:21 csharp ok
11:22 csharp hmm - I wonder why this is just now coming up...
11:22 jeff actor.usr.expire_date / "Privilege Expiration Date" deals with circulation/checkout privileges.
11:30 kmlussier Can an expired account view patron information?
11:30 tsbere expired accounts appear to be solely for circ/holds purposes. The rest of the system has no concept of expired.
11:35 ericar_ joined #evergreen
11:37 jeffdavis joined #evergreen
11:38 Bmagic joined #evergreen
11:47 jboyer-isl @bugorfeature : MARC 006
11:47 pinesol_green jboyer-isl: bugorfeature Randomly suggests whether it is a bug or feature
11:48 jboyer-isl Bah.
11:48 miker right ... you can deactivate an old staff account, or "delete" it
11:48 jboyer-isl MARC 006,
11:48 jboyer-isl @bugorfeature
11:48 pinesol_green jboyer-isl: It's a feature!
11:48 jboyer-isl Incorrect!
11:48 miker (that was directed at csharp)
11:48 eby joined #evergreen
11:50 * berick smells a "deactivate expired staff accounts" cron job already in the works
11:50 * jboyer-isl has something that can be shared if it's desired, doesn't care about staff/nonstaff though
11:59 RoganH joined #evergreen
12:12 Bmagic csharp: sorry I had to run into a meeting
12:13 jeffdavis hrm, haven't seen this error before with uploading offline transactions: "User login session has either timed out or does not exist" at /openils/var/cgi-bin/offline/offline.pl:93
12:28 Shae_ joined #evergreen
12:35 bshum jeffdavis: Well that sounds not good.
12:45 jeffdavis weirdly, the line where it fails is just trying to do open-ils.auth.session.retrieve with a valid authtoken
12:48 bshum jeffdavis: So maybe the problem is that the user that tried to initiate the upload processing timed out or something?
12:49 bshum Seems weird though.
12:52 bbqben joined #evergreen
12:53 mmorgan jeffdavis: Perhaps this is your issue? lp 1174970
12:53 pinesol_green Launchpad bug 1174970 in Evergreen "logging out of the staff client while offline transactions are being processed will cause NO_SESSION exceptions for the remaining transactions" (affected: 1, heat: 6) [Medium,Triaged] https://launchpad.net/bugs/1174970
12:56 bshum mmorgan++ # That sounds like it could be a culprit...
13:07 jeffdavis It's a good suggestion, but the session itself seems fine.  With these steps: (1) login, (2)  pull authtoken for login from logs, (3) attempt offline circ upload and get the NO_SESSION error, (4) do 'request open-ils.auth open-ils.auth.session.retrieve "AUTHTOKEN"' via srfsh and I can retrieve the session.
13:08 jeffdavis Sorry for the incoherent grammar :)
13:12 jeff jeffdavis: check your offline-config.pl to ensure that it is valid and correct, not pointing to an old opensrf_core.xml?
13:14 jeff jeffdavis: i would expect things to fail before that error if you had a missing config, but if the config is valid, but for testing and not production, or similar... the memcached instance isn't going to have your session token.
13:16 berick and make sure you don't see this near the top offline.pl:
13:16 berick do '##CONFIG##/offline-config.pl';
13:17 berick ##CONFIG## is replaced during build
13:17 * jeffdavis nods
13:18 jeffdavis offline-config.pl db settings and path to opensrf_core.xml are correct, and offline.pl has the correct path for offline-config.pl
13:18 jeffdavis I wonder if something is looking in the wrong place for memcached
13:19 berick RoganH: what do you mean that you'd favor "blocking" re: #1474051 ?
13:20 RoganH berick: blocking was probably a poor choice of words, I don't think we need to be able to save those fields in the db at all due to privacy issues
13:21 berick RoganH: gotcha.. you'd +1 dropping the columns
13:21 RoganH berick: exactly
13:21 RoganH berick: I'll update it and clarify (I'm brain dead today, think I"m fighting my wife's cold off)
13:22 berick RoganH: no problem.  thanks for clarifying
13:25 RoganH berick: on the plus side I'm writing cataloging documentation while I feel like crap so at least I'm not ruining a good day by doing it.
13:27 berick RoganH: haha.
13:27 berick way to stay positive
13:28 RoganH https://www.youtube.com/watch?v=L2Wx230gYJw
13:28 RoganH always look on the bright side on life ....
13:54 collum joined #evergreen
13:59 dbwells jeffdavis: If you have INFO level logging on, you should see line in osrfsys.log like "No user in the cache exists with key oils_auth_...".  That should at least tell you if your request is getting to the right place and whether your expected token is getting through (it will be the "..." in that phrase).
14:00 dbwells I'm wondering if the 'ses' param isn't getting through for whatever reason.
14:05 jeffdavis dbwells: Yes, I am seeing "No user in the cache exists with key oils_auth_" (i.e. with ses missing) after clicking "Upload"
14:22 ningalls joined #evergreen
14:23 dbwells jeffdavis: This might be an odd question, but what do you see for the offline GET requests in the Apache access log?  Is 'ses' missing there, too?  (If not, maybe this is a quirk with the Apache config.)
14:24 jeffdavis ses is present and correct on the GET requests.
14:29 jeffdavis hm, no, I take it back. When I click Upload, there is a GET /cgi-bin/offline/offline.pl which is missing cgi params, although they are present for other requests to offline.pl
14:33 jeffdavis When it works on a test server, I see GET requests to upload_xacts.xhtml and upload_xacts.js, then POST /cgi-bin/offline/offline.pl, then  GET /cgi-bin/offline/offline.pl with cgi params.
14:33 jeffdavis When it fails in production, I don't see the post, and the subsequent GET /cgi-bin/offline/offline.pl is missing CGI params.
14:40 Bmagic csharp: more on the invoces: We send a purchase order for a specific provider+vendorcode. The EDI goes to Baker and Taylor with that embedded. The invoice that we get back does not contain the specific vendorcode and therefore it's up to Evergreen to decide which to link.
14:41 RoganH joined #evergreen
14:42 Bmagic csharp: presumably Evergreen has all of the information in order to link it back to correct vendor code because the invoice mentions the PO number. And the PO has the vendorcode. But on my server, it's linking it back to a different vendorcode that seemly random
14:42 Bmagic csharp: we have 10 different EDI account with the same SAN but different vendorcodes
14:47 kmlussier Bmagic: Is it possible that the invoice you're receiving is truncating either the PO number or the lineitem ID's? We had trouble with that particular vendor at one time because they were truncating the lineitem IDs
14:48 Bmagic kmlussier: I don't think we are having that issue. As far as I know, all of the information is sound but EG is matching the invoices back to the wrong EDI account (maybe because we have so many of the same SANs on different accounts)
14:49 kmlussier Bmagic: We have multiple accounts with the same SANs too.
14:49 Bmagic kmlussier: oh good! I guess you don't have this issue?
14:50 kmlussier Bmagic: Not that I've heard.
14:50 Bmagic Are you getting the vendorcode in the EDI invoice from the vendor? NAD+BY ?
14:50 kmlussier Bmagic: I'll take a look and let you know.
14:50 Bmagic kmlussier: oh thanks!
14:55 bbqben joined #evergreen
15:01 dbwells jeffdavis: So your production setup GETs upload_xacts.xhtml but never does the POST?  Or does it not GET upload_xacts.xhtml at all?  All that file does is provide a form for the matching js to post, so if it isn't posting, maybe the js is dying for some reason, though I would expect to see at least one more error message somewhere along the line.
15:03 kmlussier Bmagic: I just pulled up EDI invoices for two different libraries from this vendor, and they both have the vendorcode in the EDI message.
15:03 jeffdavis dbwells: production does GET upload_xacts.xhtml.  I'm not sure if it's not doing the POST or if that request/response is failing partway through, so that Apache doesn't log it.
15:12 dbwells jeffdavis: Do you maybe get additional error information in the staff client JavaScript Console when clicking the errant Upload button?
15:13 dbwells I'm sorry, but nothing is jumping out at me glancing through the code, or thinking of recent changes.
15:17 jeffdavis dbwells: nothing helpful in js console unfortunately.
15:17 jeffdavis dbwells: thanks for giving it some thought! seems like it will end up being some obscure local config thing - ldw thought maybe our SSL handling
15:22 dbwells jeffdavis: Were I in your place, it's at this point I'd probably start doing something desperate, like alerting or console logging xul_param('ses') where it gets set inside of upload_xacts.js, then work forward or backwards from there depending on whether it's actually set or not.  My guess is you have found a legitimate but maybe obscure bug, but who knows :)
15:24 Bmagic kmlussier: wow thats great
15:24 Bmagic kmlussier: that would explain how they are getting matched up correctly. Did you happen to check Baker and Taylor?
15:24 Bmagic or that WAS B+T ?
15:24 kmlussier Bmagic: Yup. Baker and Taylor was the vendor I checked.
15:25 Bmagic kmlussier: sweet, do you mind if I mention your library to them?
15:26 kmlussier Bmagic: Not at all
15:26 Bmagic kmlussier: they told me that they do not put the vendorcode in the EDI invoices!
15:26 dbwells jeffdavis: One more question, then I should run.  Do you get the same error with a different "offline session" and/or different file?  If the file ends up being the culprit, it could be interesting to dissect.
15:28 akilsdonk joined #evergreen
15:29 jeffdavis dbwells: Good question. Same error for multiple different users, files, and sessions.
15:30 dbwells I suspected that was checked out already, but good to verify, thanks.
15:32 dbwells jeffdavis: sorry, one more possibly obvious question, for the record:  no output for the post/upload in the Apache error log either?
15:34 jeffdavis not that I can see :)
15:35 dbwells It might also be instructive to know if files are actually making it as far as your ../offline/pending dir for these cases.  Anyway, I'm out for today, good luck!
15:41 jeffdavis Thanks!
15:49 jlitrell joined #evergreen
15:49 Bmagic kmlussier++ # EDI troubleshooting
16:01 bbqben joined #evergreen
16:05 bshum Hmm, point releases due for Wednesday.
16:05 * bshum idly speculates on what's due to be merged...
16:41 bbqben joined #evergreen
16:45 Lualaba joined #evergreen
16:46 Lualaba Hello anyone could answer what is easiest way to import marcxml records to postgres sql.
16:48 Lualaba there is sample of marc xml: http://paste.evergreen-ils.org/10
16:50 Lualaba i need just need good step by step document document
17:10 mmorgan left #evergreen
17:24 bbqben joined #evergreen
17:28 gsams apparently sipserver went insane on me today, glad I figured that out relatively quickly.
17:30 jeffdavis dbwells: ldw's idea was correct, turns out we were forcing SSL on offline.pl which broke offline upload
17:31 jeffdavis more specifically the redirect involved in forcing SSL was making it choke
17:32 gsams does anyone run a cronjob to stop and start sip on a regular basis?
17:35 dbs gsams: following the lead of _robbat2|irssi, we just made our whole VM restart everything cleanly and automatically on reboot and stopped worrying about individual services :)
17:36 dbs e.g. http://pastebin.ca/3060218
17:37 jeffdavis we do a nightly SIP restart
17:38 jeffdavis although it looks like relatively recent changes to SIPServer may make it unnecessary
17:49 miker jeffdavis: they do, indeed, seem to make it unnecessary, if you use the Multiplex personality
17:50 miker if you continue to use Prefork, restart is still recommended
17:50 miker even if not, strictly, necessary with TCP keepalive enabled
17:52 jeffdavis miker: understood, thanks. Looking forward to trying it out once I've caught up on missed sleep from this past weekend's upgrade. :)
17:54 gsams dbs: well that will probably come in handy soonish when we do our server upgrade and move to VMs.
17:55 miker jeffdavis: it's a beautiful thing to watch a spinning, rogue sip2 client that would have tipped over a 16G RAM server in 90 seconds have no appreciable effect on a 2G RAM server :)
17:55 gsams For some reason SIP appears to never get properly stopped, so I just ended up with about 90 instances+ of SIPServer running our server into the ground
18:08 kmlussier Lualaba: Is this helpful at all? http://docs.evergreen-ils.org/2.8/_mig​rating_your_bibliographic_records.html
18:16 Lualaba Thank you but i need correct example of "records_out.xml"
18:18 pastebot "lualaba" at 64.57.241.14 pasted "Please check" (69 lines) at http://paste.evergreen-ils.org/11
18:22 jeffdavis Lualaba: The MARC records you've pasted contain newlines. If you remove those, so that your XML file contains only one MARCXML record per line, you should be able to use the Postgres COPY command to copy the records into a table.
18:23 Lualaba coped successful but receive error on last step: ERROR:  relation "biblio.record_entry" does not exist LINE 1: INSERT INTO biblio.record_entry (marc, last_xact_id) VALUES ...                     ^ QUERY:  INSERT INTO biblio.record_entry (marc, last_xact_id) VALUES (stage.marc, 'IMPORT') CONTEXT:  PL/pgSQL function staging_importer() line 5 at SQL statement
18:24 jeffdavis It sounds like your database does not contain the Evergreen schemas and tables. The instructions on the page kmlussier linked to assume that you have installed Evergreen already.
18:26 Lualaba DB exist
18:26 Lualaba evergreen | evergreen | UTF8     | C           | C           |   postgres  | postgres  | UTF8     | en_US.UTF-8 | en_US.UTF-8 |
18:28 jeffdavis When you are connected to the database with psql, what is the output of this command:
18:28 jeffdavis \d biblio.record_entry
18:29 Lualaba id           | bigint                   | not null default nextval('biblio.record_entry_id_seq'::regclass)  creator      | integer                  | not null default 1  editor       | integer                  | not null default 1
18:30 Lualaba may be i should first select DB after try?
18:30 Lualaba \c evergreen?
18:33 jeffdavis If you're using the command in step 1 (psql -U <user-name> -h <hostname> -d <database>), you should already be connected to the correct db.
18:33 Lualaba Good error now: ERROR:  Attempted to INSERT MARCXML that is not well formed CONTEXT:  SQL statement "INSERT INTO biblio.record_entry (marc, last_xact_id) VALUES (stage.marc, 'IMPORT')" PL/pgSQL function staging_importer() line 5 at SQL statement
18:36 jeffdavis The biblio.record_entry table has a trigger "a_marcxml_is_well_formed" which rejects inserts/updates if the MARCXML is not well formed.
18:36 bbqben joined #evergreen
18:37 jeffdavis Your options are to clean up your MARCXML so that the trigger accepts it, or else disable that trigger (but once you re-enable it, you won't be able to save changes to your records unless you clean them up).
18:37 Lualaba Thank you can you give me one example or what is wrong in my record new lines?
18:38 Lualaba or where is restrictions for: a_marcxml_is_well_formed
18:38 jeffdavis well, this one is missing a closing </record> tag: http://paste.evergreen-ils.org/10
18:39 Lualaba this is new: http://paste.evergreen-ils.org/11
18:39 jeffdavis "well formed" basically means that your XML doesn't contain syntax errors
18:39 Lualaba i generate it with perl
18:40 Lualaba i will check my xml. Thank you my friend
18:40 jeffdavis good luck!
18:42 Lualaba after i import records thats all or i need make some additional steps?
18:42 jeffdavis joined #evergreen
18:44 jeffdavis It depends what you're trying to do.
18:45 jeffdavis If you are migrating a library, you will need to add holdings (volumes and copies) to your records once they are imported.
18:46 jeffdavis I would suggest looking over the documentation on "Migrating from a legacy system" at http://docs.evergreen-ils.org/2.8/_​migrating_from_a_legacy_system.html - the page that kmlussier provided earlier is one part of that.
19:10 dcook joined #evergreen
20:08 bbqben joined #evergreen
20:27 finnx joined #evergreen
22:43 jihpringle joined #evergreen

| Channels | #evergreen index | Today | | Search | Google Search | Plain-Text | summary | Join Webchat