Evergreen ILS Website

IRC log for #evergreen, 2018-01-12

| 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
06:31 pinesol_green News from qatests: Test Failure <http://testing.evergreen-ils.org/~live>
07:46 rlefaive joined #evergreen
08:58 bos20k joined #evergreen
09:07 Dyrcona joined #evergreen
09:08 rjackson-home joined #evergreen
09:22 yboston joined #evergreen
09:27 kmlussier joined #evergreen
09:42 terran joined #evergreen
09:50 Dyrcona Some might find this interesting: https://blog.2ndquadrant.co​m/pg-phriday-postgres-zfs/
09:50 Dyrcona Makes me happier I choose ZFS on NVMe for our new db server. :)
10:00 terran Good morning. Has anyone here seen Hatch working on a 32-bit version of Windows 7? I don't have access to a 32 bit Win 7 machine, but we're getting some reports from the field.
10:11 Dyrcona terran: Are those reports negative?
10:20 terran Dyrcona: Ha, yes. They're saying they're getting a white screen that they're unable to clear up with the instructions we've given them. I'm not 100% sure they have everything installed correctly, but since some of the same people have managed to get their 64 bit machines working, I'm just wondering if there is a 32 bit problem.
10:29 jvwoolf joined #evergreen
10:33 Dyrcona I thought I saw something about a white screen recently (not necessarily with Hatch), but I'm not finding it.
10:33 Dyrcona It was either Hatch or the web client doing a white screen.
10:33 Jboyer joined #evergreen
10:34 Jboyer *poof*
10:34 Jboyer I was just looking over the logs and noticed terran's Hatch question.
10:35 Christineb joined #evergreen
10:35 Jboyer Do you know if they're trying to install Hatch 0.1.3 or 0.1.5? Nothing else will work on 32 bit, and 0.1.5 fixes a permissions issue that may also help.
10:36 csharp Jboyer: you mean 0.1.4?
10:36 Jboyer I do.
10:36 * csharp remembers a Monty Python-esque numbering confusion
10:37 csharp "1, 2, 5.  3, sir!"
10:37 Jboyer Seems like a good Four Yorkshiremen sketch could come out of it too with enough work.
10:38 csharp Jboyer: we are having them install 0.1.4 - we have no idea what they *actually* have installed, of course - terran  is asking for screenshots
10:38 terran JBoyer: They've said they are, but I've asked for a screenshot.
10:38 terran Jinx.
10:38 csharp yeah, that^^
10:38 Jboyer When is your upgrade, anyway? tomorrow night, sooner?
10:38 csharp tomorrow night
10:38 Jboyer Good luck!
10:38 terran Thanks!
10:39 csharp Jboyer: thank you
10:39 csharp we've prepared everyone as best we can for the web client - if it all falls apart it won't be for lack of effort :-)
10:40 Dyrcona I've got a question not related to the current conversation.
10:41 Jboyer Maybe with some more people using it more regularly we can smooth things out for some of our libraries.
10:41 Dyrcona Has anyone seen the Circulation > Browse Hold Shelf load more slowly for one library versus another when the latter even has more items on the hold shelf?
10:41 Jboyer we have 1 that has never used the XUL client so I've prioritized web fixes a LOT higher than I normally would. (we haven't moved them over to the XUL client because reasons...)
10:42 Jboyer Dyrcona: I haven't heard of anything like it, no. I don't know how often that interface is used here.
10:43 Dyrcona I've had a report of it yesterday, and I verified it myself. I'm told it's continuing today.
10:43 jeff Dyrcona: xul client? how much more slowly? do you have a comparison for two libs that includes number of items on hold shelf and time to load for each?
10:44 jeff Dyrcona: and for your verification, you were able to reproduce the issue from a workstation at a location where the primary difference was the library whose hold shelf you were browsing?
10:44 Dyrcona jeff: Yes, and I could time it. Yesterday, library a had 120 items and took about a minute, library had 400 and took a few seconds.
10:44 jeff (as opposed to each library loading from their location on their workstation, etc)
10:44 Dyrcona Yeah, I verified the difference from my own workstation.
10:44 jeff and since you were verifying, it's not something like each library measuring something different.
10:45 jeff hrm.
10:45 Dyrcona I'll try again today with a timer running.
10:45 jeff was there a difference between the two in terms of number of items on the hold shelf that were cancelled, or shelf-expired, etc?
10:46 Dyrcona I'll get those numbers.
10:46 jeff i am intrigued but probably shouldn't wander too far down this rabbit hole today. maybe this weekend. :-)
10:47 Dyrcona :)
10:48 Dyrcona Took about 10 seconds for library B with 448 items.
10:49 Dyrcona Library B has 2 Canceled and 446 Ready for pickup.
10:50 Bmagic csharp: do you have something in place to kill long running queries. I was thinking last night that I should advise you to get something in place because of bug 1738488
10:50 pinesol_green Launchpad bug 1738488 in Evergreen 3.0 "Web client: patron billing history results in long running query" [Medium,Confirmed] https://launchpad.net/bugs/1738488
10:51 Bmagic If you don't have something to kill those off, your database could easily get clobbered once hundreds of staff are using the Web client.
10:51 Bmagic of course, that's not the real* fix
10:52 Dyrcona I've had to kill a number of those by hand in testing and training.
10:52 Dyrcona I've noticed that they don't run for days with join_collapse_limit at 9 in production on 2.12.
10:52 Dyrcona It's on the bug, but just throwing that out there.
10:53 Dyrcona jeff: So, library A loaded in about 10 seconds, also. I'm not seeing a slow down today.
10:53 jeff drat.
10:54 Dyrcona Actually, more like 7 to 8 seconds.
10:54 Dyrcona I didn't stop the stop watch soon enough, but waited until details filled in.
10:54 jeff (or the opposite of drat, depending on your chosen perspective)
10:54 Dyrcona yes. I'm still going look at hold status. That seems like a decent angle of attack.
10:55 Dyrcona Library A has 272 this morning.
10:58 csharp Bmagic: we have a cron that kills off long-running opac searches
10:59 csharp I'll keep an eye out for long-running ones when we go live on Tuesday
10:59 Bmagic csharp: you might want to expand that to look for queries that look like ~ SELECT "mp".amount, "mp".id, "mp".note, "mp".payment_ts,
11:00 Dyrcona We've got a cron or nagios something that reports long running queries.
11:00 Dyrcona Just reports them.
11:00 Bmagic from what I've heard about your consortium and the sheer volume of requests, you might not have time to react!
11:00 Dyrcona So, Library A also has 18 canceled items on the hold shelf.
11:00 Dyrcona Thanks for the suggestions, jeff!
11:01 Bmagic our CPU on db1 goes over 60% everyday (until those queries are killed)
11:01 csharp Bmagic: whoa
11:02 Bmagic csharp: right! I cannot understate how important this is and especially for you (most likely)
11:02 csharp Bmagic: I really appreciate the heads up
11:03 Bmagic other than that, the upgrade for us was very smooth
11:03 csharp probably hogs DB backends until connections are exhausted
11:04 csharp good argument for pg_bouncer or similar (which we don't have installed)
11:04 Bmagic csharp: whichever solution you choose, but just be aware!
11:05 Bmagic It was alarming for me (at first)
11:05 * jeff absent-mindedly takes a sip of yesterday's coffee
11:05 csharp lol
11:05 csharp jeff: sorry :-/
11:05 * Bmagic watches jeff with a smirk
11:06 Bmagic it could have been two day old coffee....
11:06 csharp @dessert jeff
11:06 * pinesol_green grabs some Moon Pie and some RC Cola for jeff
11:08 Dyrcona Hm... One of these canceled holds is from 2013!
11:08 Dyrcona No, I take that back, a lot of them are and some from 2012....
11:09 Dyrcona Yeah, I noticed relatively high load on training and testing until those queries were killed, too.
11:10 Dyrcona Setting join_coallapse_limit may help some. I did that for other reasons on production, but I think it is helping with the payment history queries.
11:10 jeff Dyrcona: also, if you get to a point where you're suspecting the base query, -- yeah, I think you mostly already got where I was going.
11:11 Dyrcona jeff: That query does some ridiculous joins, but I won't repeat too much of what was discussed in IRC recently and linked from the bug, IIRC.
11:11 jeff That IDL view (which I think constitutes the base query for that interface) may suffer from the join-ordering issue.
11:12 jeff wait, which bug -- oh, bug 1738488?
11:12 pinesol_green Launchpad bug 1738488 in Evergreen 3.0 "Web client: patron billing history results in long running query" [Medium,Confirmed] https://launchpad.net/bugs/1738488
11:12 Dyrcona Heh.
11:12 Dyrcona Yeah, I switch gears without warning. :)
11:13 jeff fwiw, my "if you get to a point where you're suspecting the base query" line above was still regarding the previous gear -- the holdshelf issue you were looking at. :-)
11:13 Dyrcona Something else I want to do on productions is turn off full_page_writes. ZFS pretty much guarantees that happens anyway and it should give a performance boost.
11:14 Dyrcona jeff: I figured after the fact... :)
11:14 Dyrcona I'm not suspecting anything except staff and/or network at Library A.
11:17 Dyrcona Heh. Maybe someone was trying to view payments in the web staff client at the same time as someone else trying to browse the hold shelf? :)
11:31 rlefaive joined #evergreen
11:37 bshum joined #evergreen
11:37 berick joined #evergreen
11:38 maryj joined #evergreen
11:39 JBoyer joined #evergreen
12:10 JBoyer joined #evergreen
12:15 khuckins joined #evergreen
12:48 jihpringle joined #evergreen
12:56 dbs 5 is right out.
13:02 csharp dbs++
13:15 Dyrcona :)
13:50 kmlussier @dessert
13:50 * pinesol_green grabs some Chocolate Eclairs for kmlussier
13:50 pinesol_green [evergreen|Kyle Huckins] lp1742194 - Print Current Bills only printing circ bills - <http://git.evergreen-ils.org/?p=​Evergreen.git;a=commit;h=56bcda5>
14:17 Dyrcona @weather
14:17 pinesol_green Dyrcona: Methuen, MA :: Overcast :: 62F/17C | Friday: A steady rain. The rain will be heavy at times. High 61F. Winds S at 10 to 20 mph. Chance of rain 100%. Rainfall around a half an inch. Friday Night: Periods of rain. Low around 45F. Winds SSW at 10 to 15 mph. Chance of rain 100%. Rainfall may reach one inch. Rain and snow melt may lead to flooding.
14:18 Dyrcona Crazy....
14:21 terran We might get snow flurries.
14:25 kmlussier Dyrcona: Just wait until tomorrow when temps drop and it all turns to ice.
14:26 Dyrcona Yeah, Accuweather says high tomorrow 53 degrees Fahrenheit with a low of 13...
14:55 troy__ if midwest, we all know midwest doesn't play by any rules :)
14:56 troy__ 55 yesterday for MI was well out of the norm, but today 2-5 inches of snow
14:59 rlefaive joined #evergreen
15:33 rlefaive joined #evergreen
15:39 troy__ question: in a custom script we're calling open-ils.circ.renew, which uses our global admin user to auth (no workstation). this, however, is ignoring our circ policies for renewal and renewing all items for the same periods. is there a setting to fix this so it follows our item renewal policies? is there a way to set a standard circ lib so that renewals get the proper renewal time?
15:40 Dyrcona troy__: Easiest way is to use a user account with a workstation.
15:45 troy__ no way to just set a global for org_unit default?
15:46 Dyrcona No. There's no global or org_unit setting.
15:47 Dyrcona How are you doing the renewal? Are you using a circ id or a circ object? (IIRC, both work....)
15:49 Dyrcona You could probably get really down and dirty and instantiate an OpenILS::Circ::Circulate::Circulator object and set the circ_lib and other fields and then call do_renew().
15:50 troy__ circ id
15:50 troy__ we're doing this via php
15:51 Dyrcona Then, you want to stick the open-ils.circ calls. I really wouldn't recommend getting dirty unless you're really comfortable with the Perl code.
15:52 troy__ yea, only one of us is familiar with perl
15:52 troy__ rest of us can just read
15:52 phasefx troy__: any reason to not use a workstation with the login?
15:53 Dyrcona Well, you'd really want to be familiar with Circulate.pm, not just Perl in general. :)
15:53 troy__ i've already spent hours of my life reading through that file :P
15:53 Dyrcona I'm looking at the run_method to see if you can use a circ object, and if you can set the circ_lib somehow.
15:54 phasefx alternately, you could probably change the home_ou of the admin user, I think that might get used in leu of workstation_lib if there is no workstation, for circ stuff
15:54 troy__ phasefx: i'm still learning a lot of this. can you set a default workstation for a user?
15:54 phasefx workstation would be better, though
15:55 phasefx troy__: whatever you do when you get an authtoken, you should be able to specify a workstation in that same step
15:55 Dyrcona phasefx is correct, however..... you can force a circ_lib, I think.
15:55 phasefx but I'm not familiar with any php libraries for this
15:56 troy__ we built our own
15:56 phasefx better to do the right thing up front than play whack-a-mole, IMO
15:56 Dyrcona If your args look like: {'circ': <circ_id:int>, 'circ_lib': <org_unit_id:int>} in JSON, it should do the renewal at the specified circ_lib.
15:57 troy__ so can open-ils.circ.renew take that extra param?
15:58 Dyrcona But, yeah, when I wrote a script to autorenew circulations, I used an account that logged in with a specific workstation to get certain rules.
15:58 troy__ ('open-ils.circ.renew', [$token, ['patron_id' => $pnum, 'copyid' => $copyid]]);
15:58 Dyrcona troy_: Yes.
15:59 Dyrcona Renew goes through run_method, and the args hash is passed to the circulator, so you can set some extra Circulator fields in the args hash.
16:00 troy__ thank you so much
16:00 troy__ my basic perl self only looked at notes for what it could take
16:00 troy__ Dyrcona++
16:00 troy__ phasefx++
16:01 Dyrcona Caveat is, I've not actually tried passing circ_lib, but it looks like it should work from the code, and I've passed similar arguments before, like opac_renewal, so...
16:01 troy__ well we're gonna find out!
16:01 troy__ will make patrons much happier if this works
16:05 Dyrcona troy__: Not sure what you're doing, but it sounds like an auto-renew script? You might want to try the opac_renewal option and see how that works for you.
16:06 troy__ we built an api for our website
16:06 Dyrcona That *should* make it look more or less like a patron renewed it through the OPAC.
16:06 troy__ so patrons can hit renew on their items
16:07 Dyrcona If you're running the server with Apache, you can use an auth module to have patrons login first, and you might be able to use their authtoken.
16:08 * Dyrcona searches config for the module name.
16:09 Dyrcona Y
16:09 Dyrcona OpenILS::WWW::Proxy::Authen
16:10 Dyrcona It requires ModPerl on the server and might require some bits of OpenSRF and Evergreen.
16:11 troy__ our api server is on nginx, but our evg server is apache
16:11 Dyrcona Yeah, it requires OpenSRF and the ability to talk to the public router, at least.
16:11 troy__ k, we'll dig more
16:11 Dyrcona Well, then, it won't work for you.
16:12 Dyrcona I've used it before for a thing for staff.
16:14 * Dyrcona recalls getting the nutty idea that an OAuth provider would be a cool thing to have some time ago. :)
16:14 Dyrcona OAuth2 at any rate.
16:16 troy__ i greatly appreciate your help Dyrcona
16:16 Dyrcona YW.
16:16 troy__ Dyrcona++
16:16 troy__ Dyrcona++
16:25 kmlussier @praise Dyrcona
16:25 * pinesol_green Dyrcona LOVES the RESISTANCE!
16:29 Dyrcona heh
16:54 kmlussier OK, I'm out. Have a nice weekend everyone!
17:18 bshum joined #evergreen
17:26 jvwoolf left #evergreen
18:31 pinesol_green News from qatests: Test Failure <http://testing.evergreen-ils.org/~live>
18:59 alynn26_ joined #evergreen
19:05 alynn26_ joined #evergreen

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