Evergreen ILS Website

IRC log for #evergreen, 2014-12-03

| 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
01:16 mike_pizza joined #evergreen
01:16 mike_pizza hi
01:55 akilsdonk_ joined #evergreen
02:13 nhilton joined #evergreen
03:51 arthurdent joined #evergreen
05:55 akilsdonk joined #evergreen
06:03 wsmoak joined #evergreen
06:03 wsmoak joined #evergreen
07:48 collum joined #evergreen
07:53 rjackson-isl joined #evergreen
07:55 akilsdonk_ joined #evergreen
07:59 sarabee joined #evergreen
08:19 ericar joined #evergreen
08:21 jboyer-isl joined #evergreen
08:36 akilsdonk joined #evergreen
08:39 mmorgan joined #evergreen
08:43 mrpeters joined #evergreen
09:28 yboston joined #evergreen
10:05 loocorez_ joined #evergreen
10:12 * mmorgan is trying to understand entries like this in the pg logs for catalog searches:
10:12 mmorgan 2014-12-02 11:28:57 db postgres[19609]: [93-1] 2014-12-02 11:28:57.262 EST [172.29.120.66(43932)] [19609]: [92-1] user=evergreen,db=evergreen,e=00000 LOG:  duration: 81242.365 ms  statement:         SELECT* -- bib search: #CD_documentLength #CD_meanHarmonic #CD_uniqueWords #staff core_limit(10000) limit(1000) estimation_strategy(inclusion)  keyword: #descending sort(pubdate) filter_group_entry(4) filter_group_entry(4) ko
10:13 mmorgan Am I reading correctly that this search took 81 seconds?
10:14 berick mmorgan: yes
10:16 mmorgan Hmm. So this would likely time out on the user side, showing them no hits?
10:17 berick yes
10:17 berick it would time out around 60 seconds
10:19 mrpeters morning all, I'm tinkering around with bug 1361266 and am wondering what i might be missing that adding a <input dojoType="dijit.form.DateTextBox" id="dob" name="date_of_birth" required="true" /> on eg/opac/register.tt2 doesn't actually provide the dojo date picker.  I know register.tt2 includes base.tt2, which should include js.tt2 (which I think implements the dojo javascript) so I thought this might work.  Obviously not a fix for the bug, but tryi
10:19 pinesol_green Launchpad bug 1361266 in Evergreen "Patron self-registration form accepts date of birth in wrong format" (affected: 1, heat: 6) [Undecided,Confirmed] https://launchpad.net/bugs/1361266
10:20 tsbere mrpeters: Is dojo being loaded? Because dojo only loads if "wanted"
10:20 mrpeters tsbere: i think that is the question I am asking.  How do I make it "wanted" for this particular template?
10:21 mmorgan berick: I am also seeing some searches that took about 350 seconds, almost 6 minutes, which seems unreasonably long :-(
10:22 tsbere mrpeters: Er.....add a [%- want_dojo = 1 %] somewhere?
10:22 RoganH joined #evergreen
10:23 tsbere mrpeters: want_dojo gets enabled in various circumstances, usually by header.tt2, but is used in a wrapper. As such, if the page sets it, the wrapper should see it.
10:23 mmorgan SELECT  * -- bib search: #CD_documentLength #CD_meanHarmonic #CD_uniqueWords core_limit(10000) limit(1000) estimation_strategy(inclusion)  keyword: filter_group_entry(4) history site(MIDDLESEX) depth(1) is an example, not sure why this would take that long.
10:24 mrpeters ok, i see some want_dojo = 1; stuff in header.tt2
10:25 tsbere fun, we say "we always want it", then say "if we have autosuggest turned on, or google books turned on, or novelist turned on, we want it"
10:25 * tsbere thinks that "we always want it" should be moved to the advanced search page
10:26 mrpeters wait, so if we always say we want it why wouldnt it be enabled on register.tt2?
10:26 tsbere mrpeters: It is also possible you just don't have your syntax correct. I am not an expert. :P
10:27 tsbere Or maybe we don't have *enough* of dojo in the tpac for the date picker.
10:27 mrpeters so the staff client tt2 templates load a different amount of dojo?
10:29 tsbere mrpeters: It may be that we don't have the full dijit code loaded as we aren't using most of that functionality in tpac right now
10:30 mrpeters ok, i was trying not to introduce new javascript but maybe i just need to throw a popup if it doesnt meet the required format or something
10:31 tsbere mrpeters: Specifically, have you added any javascript box that does a dojo.require('dijit.form.DateTextBox'); ?
10:32 jeff using full paths to refer to things may help here. if you're talking about Open-ILS/src/templates/actor/user/register.tt2 i believe that is wrapped by Open-ILS/src/templates/base.tt2 (and not the OPAC base, etc)
10:32 tsbere jeff: I believe he is talking the tpac self-reg form, which does use the OPAC base
10:33 jeff but if you're talking about Open-ILS/src/templates/opac/register.tt2, that does use Open-ILS/src/templates/opac/parts/base.tt2 and friends.
10:33 mrpeters jeff -- I am talking about the latter --- the /eg/opac/register page for self registration
10:34 jeff hopefully that helps point you in a good direction as far as understanding why different dojo is present in "staff only" tt2 interfaces like the user editor vs opac interfaces
10:34 mrpeters right now i've just added the want dojo, and then <input dojoType="dijit.form.DateTextBox" id="dob" name="date_of_birth" required="true" /> before the submit buttons
10:34 mrpeters it does jeff, thank you
10:37 mrpeters i imagine adding my own dateValidation.js to the TPAC would be frowned upon for core?
10:37 mrpeters http://pastie.org/9758363 type of thign
10:38 tsbere mrpeters: I would object to that specific code. If only due to lack of translation on the alert. :P
10:38 mrpeters true
10:38 tsbere mrpeters: For the date picker, you should try adding the dojo.require javascript statement somewhere.
10:43 mrpeters ok, yeah, i see how that is happening for autosuggest now
10:56 bshum I thought want_dojo was always required now
10:56 bshum Or at least, I think we did that in 2.6
10:56 bshum And up
10:57 vlewis joined #evergreen
10:58 mrpeters i think ive got myself all kinds of confused now
10:59 mrpeters I have an awesome button that does nothing and no field ha - https://eg-master.emeralddata.net/eg/opac/register
11:00 akilsdonk joined #evergreen
11:00 mrpeters im sure i need to define jsId=setDateOfBirth still though
11:12 csharp mmorgan: are you able to run those queries directly on the DB? perhaps an EXPLAIN or EXPLAIN ANALYZE would provide clues
11:13 bshum Probably have to capture more of the query string, if it's only getting the first 100 characters or whatnot
11:13 csharp mmorgan: another possibility is the need for a VACUUM ANALYZE on your metabib. tables
11:13 sarabee joined #evergreen
11:14 csharp bshum: right
11:14 bshum csharp++ # good ideas
11:14 bshum :D
11:15 mmorgan csharp: bshum: Yes, we're planning on examining those full queries. We did a vacuum analyze over Thanksgiving, and it shaved a few seconds off our searches, but didn't make them appreciably faster.
11:15 csharp mmorgan: what version are you running?
11:16 mmorgan 2.6.2
11:16 * csharp is worried about search speed moving from 2.5 to 2.7.1 next month
11:16 tsbere csharp/mmorgan: Grab everything after "bib search:" and dump it into query_parser.pl as the -query="blah" param to possibly get the full core query - That won't help with the wrapper function, though.
11:16 bshum mmorgan: And what's in "filter_group_entry(4)"
11:17 bshum "history" seems like a fairly broad search
11:17 tsbere csharp/mmorgan: Oh, query_parser.pl would be in Open-ILS/src/support-scripts/test-scripts
11:18 cokesme joined #evergreen
11:18 bshum Oooh, fancy, tsbere++
11:18 mmorgan "filter_group_entry(4)" is a format filter "Books(All)" - we are suspicious that might be part of it
11:19 cokesme left #evergreen
11:19 mmorgan "history" IS a broad search, but, hey, it happens :)
11:19 bshum mmorgan: Hey, fwiw, my metric for broad in our DB is doing searches for "Connecticut" or "Cat" and seeing how long it takes.
11:19 mmorgan and it looks a little funny when the catalog says we don't have any books on "history" ;-)
11:20 bshum What about behind the scenes, which version of PostgreSQL are you using?
11:21 mmorgan looks like PostgreSQL 9.1.9
11:21 csharp tsbere++
11:22 dreuther_ joined #evergreen
11:23 bshum mmorgan: Have you done anything with https://bugs.launchpad.net/evergreen/+bug/1374091 locally on your system?
11:23 pinesol_green Launchpad bug 1374091 in Evergreen 2.7 "unapi.bre and slow views" (affected: 1, heat: 8) [Undecided,Fix released]
11:24 bshum That was applied in 2.6.4-ish
11:24 bshum But I know that bug was a huge stumbler for our performance when we moved up to 2.6 during the summer.
11:24 bshum Along with tuning in our PG database, but that was the worst part.
11:26 mmorgan bshum: I think we did, but will need to verify.
11:28 bshum With the changes to formats in 2.6, I'd also be curious to know how your filter_group_entry pointed back
11:28 bshum And maybe something isn't quite optimal there
11:29 bshum We don't use that feature in our catalogs, so I've never tested it.
11:31 bshum csharp: I'd say speed and performance considerations are always a critical element with every upgrade.  :)
11:32 * bshum generally loves the new icons and formats, but does think it has been an interesting road walked thus far.
11:37 mmorgan So we have a few leads to follow now, thanks all!
11:37 mmorgan berick++ csharp++ bshum++ tsbere++
11:38 * mmorgan needs to investigate system crankiness.
11:39 mllewellyn joined #evergreen
11:39 sandbergja joined #evergreen
11:52 dreuther joined #evergreen
12:01 sarabee joined #evergreen
12:19 jihpringle joined #evergreen
12:21 nhilton joined #evergreen
12:23 nhilton_ joined #evergreen
12:26 ericar joined #evergreen
13:55 nhilton joined #evergreen
14:00 collum joined #evergreen
14:06 kmlussier joined #evergreen
14:17 kmlussier Good afternoon #evergreen!
14:18 * bshum waves over at kmlussier
14:42 jeff afternoon!
14:44 maryj Good afternoon!
14:45 kmlussier @dessert [someone]
14:45 * pinesol_green grabs some Apple Crisp for artunit
14:45 kmlussier artunit is well deserving of an afternoon snack.
14:49 buzzy joined #evergreen
14:57 akilsdonk joined #evergreen
15:06 nhilton joined #evergreen
15:10 Dyrcona joined #evergreen
15:10 dreuther_ joined #evergreen
15:38 phasefx berick: is it always port 7682 that the web client will try to use for osrf-websocket-translator?
15:40 berick phasefx: yes.
15:40 berick yes, we never use the non-ssl port
15:41 berick at least not in the browser client
15:41 phasefx berick: gracias; that was my next question, whether it was using SSL or something :D
15:41 berick yeah, 7680 is non-ssl, 7682 is ssl
15:42 berick for the default opensrf websocket config
15:43 phasefx berick: does it look enough like HTTP you think that I should try ListenHTTPS with pound?
15:45 berick phasefx: probably worth a shot.  i believe it looks (more or less) like a prolonged HTTPS request
15:46 phasefx wikipedia says it gets interpreted as an Upgrade request, which I'm not familiar with.  thanks man
15:48 jeff with the disclaimer that i've not tried it, you might have better luck with a ListenHTTP directive.
15:48 phasefx jeff: roger that
15:50 phasefx bigger picture, what is actually listening on that port, apache or something else?  I went through http://evergreen-ils.org/~eric​kson/browser-client-deps.html, but I think I'm missing something
15:50 jcamins phasefx: I know nothing about pound, but I do know that setting up reverse proxies with websockets is kind of weird because you have to forward the connection upgrade request.
15:51 bshum phasefx: I think apache-websockets is covered in the new OpenSRF master / 2.4 series
15:51 phasefx jcamins: sounds scary enough that I'm going to stay away from it for a while; can just work through vpn
15:51 jeff phasefx: having done a tiny about of further reading, i don't think you're likely to have pound work. depending on your needs, nginx will probably work better.
15:51 bshum Whenever 2.4.0 for OpenSRF is cut, it should have that in the new steps
15:51 phasefx bshum: gracias; I didn't check to see how old my opensrf is
15:51 berick phasefx: apache listens on the port
15:52 jcamins ... uh-oh.
15:52 berick then hands the request to a module.  for apache, it's basically a long-running request
15:52 * jcamins may have forgotten about this when he set up a reverse proxy last weekend.
15:52 jcamins Ummm... oops.
15:53 phasefx berick: there's nothing for that yet in the example config?
15:54 nhilton_ joined #evergreen
15:54 jcamins phasefx: it's probably good I saw you mentioning websockets and reverse proxies.
15:54 jeff heh
15:54 jcamins I guess I know what I'm doing tonight.
15:54 phasefx :D
15:55 berick phasefx: see opensrf master README
15:55 phasefx berick: gracias
15:56 * phasefx bumped the node.js version in the qa tests, incidentally
15:57 akilsdonk_ joined #evergreen
16:09 bshum tsbere++ # helping me read the closed_dates.js bits
16:26 arthurdent left #evergreen
16:27 mike_pizza left #evergreen
16:35 phasefx so close.. :)   WebSocket connection to blah failed: WebSocket opening handshake was canceled    Uncaught Error: WebSocket Error [object Event] : undefined   socket.onerror   closing websocket
16:36 * phasefx goes poking at opensrf_ws.js, see if he can get a better error
16:43 berick phasefx: testing in chrome?
16:43 phasefx berick: yes, chrome and chromium
16:44 phasefx happens in firefox as well
16:44 berick what happens if you go to https://hostname:7682
16:44 berick note: FF won't work with self-signed certs :(
16:45 phasefx untrusted cert.. oy.  I'm using the IP address to get around not wanting to fool with pound
16:45 berick heh, just had a spider slowly drop down on its web between me and my screen
16:46 berick in chrome, can you click through the cert problem?
16:46 kmlussier berick: And you're still sitting at the computer and typing on your keyboard?
16:46 * kmlussier would have run away screaming. :)
16:46 phasefx berick: actually got it working in firefox, added an exception for the cert
16:47 berick phasefx: oh, good.  wonder why that didn't work for me before.. *shrug*
16:47 berick kmlussier: i gently pusehd it aside with a stack of paper ;)
16:47 berick sent it on its merry way
16:48 phasefx berick: something strange though.. I tried https://foo/eg/staff/, and it gets a 404 in chrome and works in Firefox
16:48 phasefx instead of http://foo/eg/staff/
16:49 * eeevil kills timezone offset problem for renew on webby for great justice
16:49 phasefx well now that breaks too in chrome.. what did I do? *8)
16:49 berick phasefx: what you always do ;)
16:49 phasefx berick: trailing slash or lack thereof
16:49 berick @who has chaos powers
16:49 pinesol_green chatley has chaos powers.
16:49 * phasefx even broke pinesol_green ;)
16:50 phasefx berick: working in chrome for me now too
16:50 berick awesome
16:50 dreuther joined #evergreen
16:51 phasefx berick: I think starting from the non-SSL login page was a contributing factor to breakage
16:52 berick yeah, cuz then you get no chance to click through the warning
16:57 eeevil berick: heads up! I'm pulling your fixes into collab/miker/web-client-sprin​t1-bug-fixing-rebased-collab, just to keep the bug fixes together
16:57 berick eeevil++
16:58 akilsdonk_ joined #evergreen
16:59 eeevil we can squash/reorder/etc later, I figure
17:01 jeff togetherness!
17:01 * jeff grins
17:02 pinesol_green Incoming from qatests: Test Success - http://testing.evergreen-ils.org/~live/test.html <http://testing.evergreen-ils.org/~live/test.html>
17:07 mmorgan left #evergreen
17:15 mrpeters left #evergreen
17:44 nhilton joined #evergreen
21:02 sarabee joined #evergreen

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