2008-07-13T00:04:01 i give up; i can't find a url for the root of this in scrollback. tim who, and what did he say/do? 2008-07-13T00:05:18 sboyette: it's from the web4lib mailing list 2008-07-13T00:05:19 sec 2008-07-13T00:10:32 http://lists.webjunction.org/wjlists/web4lib/2008-July/047897.html 2008-07-13T00:12:44 in or abouts that thread is what dbs dbs was referring to 2008-07-13T00:18:59 i had no idea SQL was made of rich, Corinthian leather and pockets of air 2008-07-13T00:19:46 i thought it was made of the souls of old-line IBM engineers, trapped in the circuitry of a System/370 machine 2008-07-13T00:21:45 they upholster the 370's with leather ... like Cray benches, but stitched by hand 2008-07-13T00:22:44 ooooooooo 2008-07-13T00:22:46 sexy! 2008-07-13T00:23:05 i want one bank manager chair style, with the big buttons and quilting 2008-07-13T00:23:22 and a candelabra on top 2008-07-13T00:23:37 (that part doesn't have anythiing to do with chairs) 2008-07-13T00:24:22 if the candelabra can be attached with duct tape, you have yourself a deal 2008-07-13T04:07:18 *** phasefx has quit IRC 2008-07-13T04:07:18 *** peanutb has quit IRC 2008-07-13T04:08:22 *** phasefx has joined #openils-evergreen 2008-07-13T04:08:56 *** peanutb has joined #openils-evergreen 2008-07-13T07:29:34 *** kgs has joined #openils-evergreen 2008-07-13T08:45:30 *** mjg_ has joined #openils-evergreen 2008-07-13T12:16:12 *** kgs has quit IRC 2008-07-13T12:18:23 *** kgs has joined #openils-evergreen 2008-07-13T13:29:41 *** dbs has joined #openils-evergreen 2008-07-13T18:44:10 *** dbs has quit IRC 2008-07-13T19:05:11 *** dbs has joined #openils-evergreen 2008-07-13T20:00:43 *** asjoyner has joined #OpenILS-Evergreen 2008-07-13T20:01:41 greetings! I'm setting up Evergreen, and I'm having a bootstrapping problem. I can't login to the staff client because it's "not yet configured for the specified server." 2008-07-13T20:01:51 apparently I can't fix that w/o logging in as an administrative user... 2008-07-13T20:01:57 am I missing something obvious? :) 2008-07-13T20:03:00 asjoyner: right. that's part of the initial setup process. what are you installing this on? (on the server) I'll see if I can point you at the particular step in the instructions 2008-07-13T20:03:21 also, which version of the staff client and server software are you using? 2008-07-13T20:03:22 it's on ubuntu, if that's what you mean? 2008-07-13T20:03:29 it is 2008-07-13T20:03:44 I was working from this doc: http://open-ils.org/dokuwiki/doku.php?id=installing_evergreen_1.2_on_ubuntu_7.10 2008-07-13T20:03:55 although I'm actually installing on hoary hedgehog 2008-07-13T20:04:15 ahh... no compilation problems, then? 2008-07-13T20:04:49 only related to having initially chosen postfix 8.3 -- I'll send along a patch or two to make things fail more gracefully if you're interested. :) 2008-07-13T20:05:00 sure! 2008-07-13T20:05:27 yeah, postgres 8.2 for now. 2.0 will have full 8.3 support (major changes in full-text indexing) 2008-07-13T20:05:34 yeah, I noticed :) 2008-07-13T20:06:10 any luck on finding that doc reference? :) 2008-07-13T20:06:17 so the OPAC is working, though empty? 2008-07-13T20:06:21 yep 2008-07-13T20:06:23 (looking now) 2008-07-13T20:06:38 I can login to the OPAC (after some fiddling), but can't seem to do much w/ the administrative user 2008-07-13T20:07:01 which version of the staff client are you using? or, are you building the staff client on your ubuntu installation 2008-07-13T20:07:08 local build 2008-07-13T20:07:29 build id 20080713 :) 2008-07-13T20:08:42 and you're just cd'ing into Open-ILS/xul/staff_clint/build and running `xulrunner application.ini` 2008-07-13T20:08:50 ayup 2008-07-13T20:09:02 as per the instructions at the end of the a-fore-mentioned doc 2008-07-13T20:09:12 right 2008-07-13T20:09:12 it seems to leave off a bit early, as the "what now" question is left hanging. :) 2008-07-13T20:09:34 there seem to be good end-user docs, and good install docs, but that middle ground for bootstrapping seems a bit lacking 2008-07-13T20:10:11 so you have a dir called /openils/var/web/xul/20080713, which contains the server/ dir from Open-ILS/xul/staff_client/build/ 2008-07-13T20:10:17 indeed 2008-07-13T20:10:27 and i can see it via apache 2008-07-13T20:11:00 and the about button on the login interface shows the correct build id? 2008-07-13T20:11:16 indeed 2008-07-13T20:11:41 hrm ... very odd 2008-07-13T20:11:52 what's the issue? 2008-07-13T20:12:03 just the man we need! 2008-07-13T20:12:24 :) 2008-07-13T20:12:38 I'm having some trouble bootstrapping the staff client. 2008-07-13T20:12:40 phasefx_: unsupported version error, but everything seems to be set up per the ubuntu docs, step 29 2008-07-13T20:13:04 miker_: I don't get an unsupported version error... actual error message coming up, one sec 2008-07-13T20:13:18 ahh... well, I misunderstood then 2008-07-13T20:13:57 asjoyner: there's a permission associated registering a client with a server.. out of the box, there's only one user, admin, and that user has every permission 2008-07-13T20:14:04 I'm unable to login, presumably because once I select the server on the left side of the client, it still says "Not yet configured for the specified server" on the right hand side of the client. 2008-07-13T20:14:31 what happens when you try to login, and with which user? 2008-07-13T20:14:34 ahh. well, you can log in and it will have you register the instance of the staff client with the server 2008-07-13T20:14:37 Network/server failure. Please check your Internet connection to undefined and choose... 2008-07-13T20:14:42 with the admin user 2008-07-13T20:15:07 sounds like this isn't a bootstrapping problem as I thought, I should probably go turn up the logging to debug and try it again 2008-07-13T20:15:29 can you paste some of the console output from xulrunner into http://paste.lisp.org/new/openils-evergreen when that happens? 2008-07-13T20:15:30 connection to undefined ... that suggests that the Hostname (top left box) is empty 2008-07-13T20:15:43 I thought I had a process problem, as I wasn't sure how I could possibly register the client workstation w/o logging in 2008-07-13T20:16:12 Hostname: pearl.joyner.ws, and the test button causes Status and Version to say '200: Ok' 2008-07-13T20:16:24 (it's an internal-only hostname) 2008-07-13T20:16:29 :) 2008-07-13T20:16:51 :) 2008-07-13T20:17:11 I do get that when I try to connect a trunk staff client to a 1.2 server ... but that wouldn't be your problem :) 2008-07-13T20:17:27 okay, now that I know the admin login ought to work regardless of the client being registered, I'll go take a stab at the logs again. 2008-07-13T20:17:39 thanks for the very helpful discussion, gentlemen. :) 2008-07-13T20:18:11 btw, how formal is the patch submission process? I imagine it's documented on the website, I just hadn't taken the time to look yet. 2008-07-13T20:18:23 feel free to send logs to the list ... and console output is helpful 2008-07-13T20:18:25 for patches 2008-07-13T20:19:14 just describe what it does, send it as a diff -pur (or svn diff) to the open-ils-dev list, and attach a DCO ... for the "official" process see: 2008-07-13T20:19:30 http://open-ils.org/documentation/contributing.html 2008-07-13T20:19:58 ah I remember reading that before, the certificate of origin stuff rings a bell. 2008-07-13T20:20:16 the closer to the official process the better, but the bar is (for obvious reasons) lower for "fail gracefully" patches ;) 2008-07-13T20:20:36 I'll test, put some polish on it and send it out some time this week. 2008-07-13T20:20:47 awesome. thanks 2008-07-13T20:20:49 first step: get the local ILS up :) 2008-07-13T20:21:04 right on 2008-07-13T20:21:35 curious question, what time zone are most of the developers in? I'm assuming East Coast, as per Georgia PINES? 2008-07-13T20:21:56 indeed 2008-07-13T20:22:05 excellent. 2008-07-13T20:22:07 as well as Ontario 2008-07-13T20:22:13 and NC 2008-07-13T20:22:17 NC? 2008-07-13T20:22:23 UNC-CH, I'd guess? 2008-07-13T20:22:27 North Carolina 2008-07-13T20:22:39 well, not yet :) (but we can hope) 2008-07-13T20:22:44 haha 2008-07-13T20:22:52 one of the Equinox folks 2008-07-13T20:23:23 we have some folks from Europe hanging out in here ... asmodai for sure 2008-07-13T20:23:41 perhaps there's another NC developer in the works... I'd love to contribute if I have time more in the future, I've been loosely following the project since it's original announcement. 2008-07-13T20:24:13 great! 2008-07-13T20:24:36 my wife's a librarian, I'm a linux geek, so it's a fun mingling of interests. :) 2008-07-13T20:24:53 right on 2008-07-13T20:24:59 anyway, off to the logs. I'll touch base when I get it working, or if I get stuck again. Thanks Mike. 2008-07-13T20:25:04 np 2008-07-13T20:38:05 okay, so I don't get much interesting in the server-side log files at connect time after turning up the debugging to 4 2008-07-13T20:38:52 any error messages in the client console? 2008-07-13T20:39:05 I enter the connection info (Hostname), and the Re-Test server button indicates happiness 2008-07-13T20:39:24 I enter the username admin, password open-ils (the default as per the wiki, which does work for logging into the open and via srfsh) 2008-07-13T20:39:37 I get a cert warning because I'm using a self-signed cert (expected) 2008-07-13T20:39:41 and then I get... (paste bot, coming right up) 2008-07-13T20:40:43 asjoyner pasted "error from staff client" at http://paste.lisp.org/display/63652 2008-07-13T20:41:32 I do see entries in the logs corresponding, time-stamp wise, with that connection. This leads me to believe (contrary to the 'undefined' indication in the error) that it's indeed connecting. 2008-07-13T20:41:34 hrm... is memcache running? 2008-07-13T20:41:59 yes, and the config in opensrf.xml matches it's port and ip 2008-07-13T20:42:06 11211, 127.0.0.1 2008-07-13T20:42:10 I think "undefined" there is an error with the, erm, error message ;) 2008-07-13T20:42:18 yeah, I suspected as much. :) 2008-07-13T20:43:19 phasefx_: any ideas? I've only seen authenticate.complete fail when memcache isn't running or is mis-configured ... 2008-07-13T20:43:48 well, the first call is apparently working, .init 2008-07-13T20:43:58 I'd try opac logins and srfsh logins 2008-07-13T20:43:58 (reason being, the authenticate.init call tosses the challenge hash in there) 2008-07-13T20:44:04 both work 2008-07-13T20:44:09 phasefx_: both .... yeah :) 2008-07-13T20:44:39 asjoyner: any more interesting text in the same console in which xulrunner is invoked? 2008-07-13T20:44:43 math_bench 10 works as expected, etc. 2008-07-13T20:44:50 annotating... 2008-07-13T20:46:21 and we're not dealing with a json trunk vs json rel_1_2 issue... 2008-07-13T20:46:59 haha 2008-07-13T20:47:04 "paste too large" :) 2008-07-13T20:47:07 heh 2008-07-13T20:47:11 trimming it a bit... 2008-07-13T20:47:38 asjoyner annotated #63652 with "console output from staff client" at http://paste.lisp.org/display/63652#1 2008-07-13T20:47:41 there we go 2008-07-13T20:48:04 hopefully that's the relevant bits you're after 2008-07-13T20:48:15 if you'd like the client initialization, I'll gladly include it in a second paste 2008-07-13T20:48:55 okay, so whatever is happening, I suspect it's happening at the gateway...that ! NETWORK FAILURE is being spit out by RemoteRequest 2008-07-13T20:49:24 but 2008-07-13T20:49:27 it works with OPAC 2008-07-13T20:49:31 I have the gateway logging turned up to debug, I didn't notice anything relevant, would you like me to paste it? 2008-07-13T20:49:31 oy :) 2008-07-13T20:50:26 let me trace that code 2008-07-13T20:50:26 asjoyner annotated #63652 with "gateway logging at debug log level" at http://paste.lisp.org/display/63652#2 2008-07-13T20:50:37 asjoyner: if you can grab the section between authenticate.init and (what we'll hopefully sees as) the call to .complete ... 2008-07-13T20:50:40 that's from /openils/var/log/gateway.log 2008-07-13T20:51:13 do you see the call to complete following that? 2008-07-13T20:51:27 may be a ways down at debug level ;) 2008-07-13T20:51:27 check apache log? 2008-07-13T20:51:39 what I pasted it through the end of the log file, there is a Completed processing service=open-ils.auth 2008-07-13T20:51:42 checking... 2008-07-13T20:51:45 error.log, presumably? 2008-07-13T20:51:53 I'd look at error.log and access.log 2008-07-13T20:52:34 Completed processing service=open-ils.auth, method=open-ils.auth.authenticate.init just means the init call finished ... the .complete never got to far enough along to be logged 2008-07-13T20:52:44 or died before it could log 2008-07-13T20:53:34 asjoyner annotated #63652 with "apache's relevant error.log " at http://paste.lisp.org/display/63652#3 2008-07-13T20:53:50 presumably the Connection closed... is where the child died before returning data to Apache? 2008-07-13T20:54:04 that doesn't look promising 2008-07-13T20:54:16 access.log just shows the POST to /gateway 2008-07-13T20:54:26 after which it seems to a'splode. 2008-07-13T20:55:01 asjoyner: feel like hacking your client? 2008-07-13T20:55:09 sure 2008-07-13T20:55:09 heh 2008-07-13T20:55:17 start throwing in debug prints? :) 2008-07-13T20:55:20 actually, let's try one last thing before that ;) 2008-07-13T20:55:33 I just wanted to turn SSL off for those calls 2008-07-13T20:56:26 do you get anything back from https://pearl.joyner.ws/gateway?service=open-ils.auth&method=open-ils.auth.authenticate.complete¶m=%7B%22username%22%3A%22admin%22%2C%22password%22%3A%223b4f305d860cf59224374211e4650dab%22%2C%22type%22%3A%22temp%22%7D 2008-07-13T20:56:53 {"payload":[],"debug": "osrfMethodException : No authentication seed found. open-ils.auth.authenticate.init must be called first","status":404} 2008-07-13T20:57:00 there we go 2008-07-13T20:57:33 so, could memcache have been restarted after the opensrf services came up? 2008-07-13T20:57:58 mmm... I'm pretty sure they've been restarted since then, but I'll gladly restart them. 2008-07-13T20:58:04 shall I? 2008-07-13T20:58:06 unfortunately, the libmemcache code doesn't attempt to reconnect 2008-07-13T20:58:10 miker_: how long is that seed valid after .init? 2008-07-13T20:58:13 please ... 2008-07-13T20:58:26 phasefx_: at least 5min 2008-07-13T20:58:31 k 2008-07-13T20:58:47 but that's a good point 2008-07-13T20:59:10 it'd be close... 2008-07-13T20:59:20 or past 2008-07-13T20:59:22 restarted via stop_all .. wait... start_all 2008-07-13T20:59:29 cool 2008-07-13T20:59:37 miker_'s magic long URL still returns the same result. 2008-07-13T20:59:54 shall i give the client a try 2008-07-13T20:59:55 ? 2008-07-13T20:59:56 well, as phasefx pointed out, we may have waited too long 2008-07-13T21:00:00 sure 2008-07-13T21:00:51 same behavior 2008-07-13T21:00:54 miker_: yeah, RemoteRequest would/should have caught an exception like that 2008-07-13T21:00:56 to get that magic url, I just pasted the gobbledygook after "! NETWORK FAILURE. HTTP STATUS = " after https://pearl.joyner.ws/gateway? 2008-07-13T21:01:09 makes sense 2008-07-13T21:01:10 from the console paste 2008-07-13T21:01:45 I attempted to login to the opac, and after firefox going grey to the javascript for a long time, it eventually errored out. 2008-07-13T21:02:24 and srfsh login is fine? 2008-07-13T21:02:26 oh wait, I'm an idiot 2008-07-13T21:02:39 when we got connected to the opac earlier we changed the password. :0 2008-07-13T21:02:49 :D 2008-07-13T21:03:09 heh ... well, it shouldn't throw a network error there for a bad pw, right phasefx_? 2008-07-13T21:03:16 right 2008-07-13T21:03:31 yeah, it's still doing the same w/ the right password. 2008-07-13T21:03:35 should give a nice vague warning about possibly wrong password, username, etc 2008-07-13T21:03:55 let's try this.. edit the file chrome/content/main/constants.js where you built your client 2008-07-13T21:04:12 and add 'secure' : false to the AUTH_INIT and AUTH_COMPLETE stanzas 2008-07-13T21:04:29 which is what the opac gave... although it still gives that after the long delay, the error is "Login failed. The username or pasword provided was not valid.. yada yada" 2008-07-13T21:04:37 okay, editing now 2008-07-13T21:04:41 'AUTH_INIT' : { 'app' : 'open-ils.auth', 'method' : 'open-ils.auth.authenticate.init', 'secure' : false }, 2008-07-13T21:04:45 'AUTH_COMPLETE' : { 'app' : 'open-ils.auth', 'method' : 'open-ils.auth.authenticate.complete', 'secure' : false }, 2008-07-13T21:04:49 so should look like that when you're done 2008-07-13T21:04:54 then rebuild the client and try logging in 2008-07-13T21:05:02 working... 2008-07-13T21:07:39 use the same build_id, and I don't need to copy out the server/ dir, yes? 2008-07-13T21:08:06 right, we're just testing local code at the moment 2008-07-13T21:08:28 so ... the gateway must not be happy in the 443 vhost section of the apache config 2008-07-13T21:09:15 okay, done, same error (it seems) -- what more info can I provide yoU? 2008-07-13T21:09:41 hrmm, I'm stumped at the moment (you should revert those changes) 2008-07-13T21:10:19 heh ... I interpreted "working..." as "it's working" ;) 2008-07-13T21:10:28 brb 2008-07-13T21:11:10 phasefx_, did miker_ mean to indicate that the apache vhost config is suspect? 2008-07-13T21:11:42 asjoyner: if it worked non-ssl ( 'secure' : false ), yes ... but that's not the case 2008-07-13T21:11:52 * asjoyner nods. 2008-07-13T21:12:01 asjoyner: did you still get an SSL-splode in error.log? 2008-07-13T21:12:11 checking... 2008-07-13T21:12:19 I'm assuming 'secure' : false actually works :) 2008-07-13T21:12:34 [Sun Jul 13 21:08:15 2008] [info] [client 10.0.5.3] Connection closed to child 9 with abortive shutdown (server pearl.joyner.ws:443) 2008-07-13T21:12:53 I don't like that abortive shutdown message 2008-07-13T21:13:12 * asjoyner agrees whole-heartedly. 2008-07-13T21:13:26 I get standard shutdowns 2008-07-13T21:14:15 asjoyner: so, did we try Mike's magic URL trick with a fresh list of params immediately after a failed login attempt? 2008-07-13T21:14:33 trying that now... 2008-07-13T21:15:02 need to be quick since the init key will expire soon after the .init call 2008-07-13T21:15:12 * asjoyner nods. 2008-07-13T21:15:26 of course,right before you said that, I poked the opac again, and now firefox is unhappy... 2008-07-13T21:15:50 seems to be the same 2008-07-13T21:15:57 I'll try it again w/ a shorter time window 2008-07-13T21:16:08 asjoyner: another question, are you shutting down apache, and starting it after opensrf is started? 2008-07-13T21:16:19 mmm... no! 2008-07-13T21:16:24 that's bad 2008-07-13T21:16:30 at least I didn't after the last time 2008-07-13T21:16:47 starting it now... 2008-07-13T21:17:42 same behavior... 2008-07-13T21:18:21 different behavior from miker_'s url when it's done quickly, though 2008-07-13T21:18:31 it hangs... presumably because the app has blown up 2008-07-13T21:18:45 it causes an abortive shutdown entry in error.log? 2008-07-13T21:19:01 no log entry yet.. waiting for it, though 2008-07-13T21:19:18 and is anything eating CPU while this happens? 2008-07-13T21:19:18 eventually got: {"payload":[],"status":200} 2008-07-13T21:19:28 hrmm 2008-07-13T21:19:37 didn't check the CPU load 2008-07-13T21:19:42 i can do it again and check if it's useful 2008-07-13T21:19:49 it wasn't obscene, load's not high 2008-07-13T21:19:52 *** mjg_ has quit IRC 2008-07-13T21:19:58 that empty payload.. I wonder how RemoteRequest treats that 2008-07-13T21:20:28 hrm... I now suspect libdbdpg is not linked to libdbi 2008-07-13T21:20:42 mmm... checking w/ ldd 2008-07-13T21:20:51 asjoyner: does autogen.sh run without errors? 2008-07-13T21:21:27 /usr/local/lib/dbd/libdbdpgsql.so is not linked against anything named libdbi 2008-07-13T21:21:46 yar ... that's one problem 2008-07-13T21:22:10 miker_: I believe it did, yes, during the install. We had to troubleshoot a username being wrong in a config at first (postgresql default not changed to evergreen), after fixing that it ran successfully. 2008-07-13T21:23:14 what version of libdbi-drivers do you have? 2008-07-13T21:23:25 I read something about this on the forums, and was giving my wife, who was driving at the time the suggestion to go check that w/ ldd... after the username change fixed the problem, we never went back and checked on the linker :) 2008-07-13T21:23:34 libdbi-drivers-0.8.3-1 2008-07-13T21:23:57 time to recompile and reinstall the libs, restart services, try again? 2008-07-13T21:24:13 just libdbi-drivers, yep 2008-07-13T21:24:37 step 17 in the instructions, just remove the --enable-libdbi param 2008-07-13T21:24:40 is --enable-libdbi insufficient? 2008-07-13T21:24:45 :) 2008-07-13T21:25:02 it's ... malformed in 0.8.3, IIRC 2008-07-13T21:25:15 * asjoyner sighs. 2008-07-13T21:25:45 the build .so still isn't linked against libdbi 2008-07-13T21:26:05 s/build/built/ 2008-07-13T21:26:09 saying either enable or disable disables it ... (they know about the problem, I believe, just waiting for a new release) 2008-07-13T21:26:12 arg 2008-07-13T21:27:26 make clean before rebuilding, of course 2008-07-13T21:27:38 ah, that'd do it 2008-07-13T21:27:51 bingo 2008-07-13T21:28:04 thanks for stating the obvious. :) 2008-07-13T21:28:24 heh ... only 'cause I've done it myself :) 2008-07-13T21:29:18 okay, so checklist... stop_all, start_all, then restart apache... anything else? 2008-07-13T21:29:56 that'll do it ... well, and, as the opensrf user run 2008-07-13T21:30:07 /openils/bin/autogen.sh 2008-07-13T21:30:18 and proximity refresh 2008-07-13T21:30:32 phasefx_: indeed 2008-07-13T21:30:47 proximity refresh? 2008-07-13T21:31:10 in srfsh, request open-ils.storage open-ils.storage.actor.org_unit.refresh_proximity 2008-07-13T21:31:52 used for holds logic, proximity of org units 2008-07-13T21:32:15 given that I've never logged into the staff client, and presumably everything in the ILS is empty, is it relevant? :) 2008-07-13T21:32:17 thankfully, dbs added that to autogen.sh (-u switch), though I don't think it's in any releases yet ... 2008-07-13T21:32:28 (autogen.sh ran w/o errors) 2008-07-13T21:32:33 rock 2008-07-13T21:32:38 that's a very good sign 2008-07-13T21:33:22 ahh... -u is in 1.2.2.2 2008-07-13T21:33:39 cool 2008-07-13T21:33:49 same behavior. :( 2008-07-13T21:33:59 though, 1.2.2.3 will be forthcoming ... 2008-07-13T21:34:22 arg! you're killin' me, man! :) 2008-07-13T21:35:08 :) 2008-07-13T21:35:18 no problems installing libdbi-drivers, though? the installed version is linked as expected? 2008-07-13T21:35:30 double checking... 2008-07-13T21:35:52 $ ldd /usr/local/lib/dbd/libdbdpgsql.so | grep libdbi 2008-07-13T21:35:52 libdbi.so.0 => /usr/local/lib/libdbi.so.0 (0xb7fa3000) 2008-07-13T21:36:03 at least there's that 2008-07-13T21:36:19 presumably it's one less problem I'll encounter later. :) 2008-07-13T21:36:21 so, does logging in through srfsh work? 2008-07-13T21:36:26 checking 2008-07-13T21:37:07 Login Session: 04277d0fb061753c05d25e515f786bc0. Session timeout: 420.000000 2008-07-13T21:37:21 Request Completed Successfully 2008-07-13T21:37:41 srfsh seems happy 2008-07-13T21:37:50 yep ... ok, so now we're back to the gateway 2008-07-13T21:37:56 math_bench seems to work as expected, etc. 2008-07-13T21:37:59 what does the magic URL return? 2008-07-13T21:38:05 checking... 2008-07-13T21:38:50 {"payload":[{"ilsevent":0,"textcode":"SUCCESS","desc":" ","pid":7575,"stacktrace":"oils_auth.c:305","payload":{"authtoken":"bc3a9fac33c7c64829819acd5f4e31ef","authtime":300}}],"status":200} 2008-07-13T21:39:04 first time I've seen that... 2008-07-13T21:39:16 but same behavior from the client 2008-07-13T21:39:21 did you restart the staff client since the change? (yeah, that's success via the gateway) 2008-07-13T21:39:36 wondering about caching 2008-07-13T21:39:49 client side caching, I mean 2008-07-13T21:40:13 shouldn't be any caching with POST and/or SSL :( 2008-07-13T21:40:24 client side meaning my browser? 2008-07-13T21:40:36 shouldn't be, although I can do the requests w/ curl to verify 2008-07-13T21:40:38 well, meaning the staff client 2008-07-13T21:40:54 mm, I'm restarting it for each iteration, does it keep a cache on disk? 2008-07-13T21:40:59 asjoyner: the staff client is still giving those ! NETWORK FAILURE messages in the console? 2008-07-13T21:40:59 which is basically ffx without any browser-ish chrome 2008-07-13T21:41:02 I don't know much about xulrunner, that's new to me. 2008-07-13T21:41:23 phasefx_: yes 2008-07-13T21:41:28 -= THIS MESSAGE NOT LOGGED =- 2008-07-13T21:41:28 service=open-ils.auth&method=open-ils.auth.authenticate.complete¶m=%7B%22username%22%3A%22admin%22%2C%22password%22%3A%221d81c962a6841e92a3b505a6a7a44900%22%2C%22type%22%3A%22temp%22%7D 2008-07-13T21:42:50 well.. the OPAC uses RemoteRequest too... 2008-07-13T21:43:08 logging in through the OPAC works or does not work now? 2008-07-13T21:43:37 checking... 2008-07-13T21:43:57 works 2008-07-13T21:44:18 and the hostname being used there matches the one you're using in the client? 2008-07-13T21:44:26 yep, fqdn in both places 2008-07-13T21:44:51 * dbs wonders if asjoyner should try rm -fr ~/.openils to get rid of anything hanging around in the workstation cache/settings 2008-07-13T21:44:56 I've used the FQDN for most things, w/ a few notable exceptions such as memcached and postgresql 2008-07-13T21:44:59 now that a is asleep 2008-07-13T21:45:31 any down sides to thwaping the ~/.openils dir? 2008-07-13T21:45:49 doesn't seem like there should be. 2008-07-13T21:45:59 it's purely client-related, it will ask you to register again 2008-07-13T21:46:34 moved it out of the way (to ~/.openils.not) -- trying again 2008-07-13T21:46:41 ahh, that's safer :) 2008-07-13T21:46:44 dbs: luckily (heh) we haven't gotten that far yet :) 2008-07-13T21:46:55 same behavior 2008-07-13T21:47:19 dbs: but we've seen another instance of 0.8.3-1 and --enable-libdbi :( 2008-07-13T21:47:55 miker_: that's probably worth a note in the wiki to have people double-check w/ ldd after the build, before install. 2008-07-13T21:48:23 asjoyner: indeed 2008-07-13T21:49:22 miker_: you know what? I'm not sure that the libdbi linking thing matters 2008-07-13T21:49:40 but that's another discussion 2008-07-13T21:49:51 asjoyner: have you tried running settings-tester.pl ? 2008-07-13T21:50:22 dbs: as a data point, it didn't positively or negatively affect the current situation. Once we get things working, I'll be glad to break the link and see if it breaks anything else. 2008-07-13T21:50:26 dbs: got a link? 2008-07-13T21:50:35 as the opensrf user, run Open-ILS/src/support-scripts/settings-tester.pl 2008-07-13T21:50:36 I saw mention of it a few hours ago while surfing around on the web 2008-07-13T21:50:57 afk one sec, wife caling 2008-07-13T21:51:48 back. 2008-07-13T21:51:54 wow, that was fast :) 2008-07-13T21:51:58 I work on fixing the ILS, wife works on fixing dinner. :) 2008-07-13T21:52:02 dbs: well, the opac is logging in now, where it wasn't before ... or we think 2008-07-13T21:52:13 miker_: nope, opac was working before 2008-07-13T21:52:20 was it 2008-07-13T21:52:21 ok 2008-07-13T21:52:25 when libdbi wasn't linked in 2008-07-13T21:52:31 checking settings-tester.pl 2008-07-13T21:52:33 well, then that's not a problem ... odd, that 2008-07-13T21:53:07 may be an issue with older libc rather than libdbi 2008-07-13T21:53:24 (dbs: this is a hedgehog install, btw) 2008-07-13T21:55:38 and I started w/ postgresql-8.3, had a bit of an a'splosion, and backed up to postgresql-8.2, so I dropped the database, rebuilt, etc... but I have a pretty high confidence that I ended up at a sane config? (except it doesn't work... :) ) 2008-07-13T21:55:41 that abortive shutdown error sounds horrrribly familiar. urgh. 2008-07-13T21:55:51 * dbs tries to jar his memory 2008-07-13T21:57:25 but only in the SC ... directly (GET url in ffx) or through the opac works, as does srfsh :( (just listing status for my own sake) 2008-07-13T21:57:54 settings tester has some errors... pastebot coming up... 2008-07-13T21:58:07 miker_: yep, I've read the backscroll - and it's reaaaly familiar 2008-07-13T21:58:46 asjoyner annotated #63652 with "settings-tester.pl output" at http://paste.lisp.org/display/63652#4 2008-07-13T21:59:54 nothing serious there, unfortunately 2008-07-13T22:00:06 yeah, reporter/setup can be ignored for now 2008-07-13T22:00:10 okay 2008-07-13T22:00:43 (too bad... I got all hopeful there for a sec... at this point I'd be glad to be told "yeah, you've got the wrong database password in a config file...") 2008-07-13T22:01:03 asjoyner: a lot of people do get that :) 2008-07-13T22:01:25 I'm not at all surprised. :) 2008-07-13T22:03:58 okay, let's recap. miker_ and phasefx_ were suggesting earlier that the "! NETWORK FAILURE" errors are likely related to the gateway, but it doesn't generate any logging output, and doesn't crash 2008-07-13T22:05:21 and your staff client is from the same version (maybe even the same directory) of Evergreen that has been installed? 2008-07-13T22:06:03 the staff client is a mozilla framework app(?) which is mostly javascript(?) and talks https to the server, but we believe it's getting back a truncated response, presumably because of the abortive shutdown in apache's error.log, presumably because something apache was calling, died? 2008-07-13T22:06:32 phasefx_: yes, the staff client is from the same directory that the server code was installed from, this is a test box with nothing on it but the single evergreen instance 2008-07-13T22:06:33 oh, that's a question, what version of xulrunner are you using? 2008-07-13T22:06:54 1.8.1.13+nobinonly-0ubuntu1 XUL + XPCOM application runner 2008-07-13T22:06:59 latest out of hoary 2008-07-13T22:08:06 that should still work 2008-07-13T22:08:10 asjoyner: ahhh 2008-07-13T22:08:19 ahhh? 2008-07-13T22:08:19 try a2dismod deflate 2008-07-13T22:08:30 and stop/start apache 2008-07-13T22:08:55 done, testing 2008-07-13T22:09:26 same behavior 2008-07-13T22:09:52 only did force-reload of apache... restarting for kicks... 2008-07-13T22:10:06 (i'm just reading through the ubuntu install docs and noting things that I do not do, and which cause other problems - mod_deflate breaks supercat in trunk, for example) 2008-07-13T22:10:20 supercat? 2008-07-13T22:10:48 dbs: non-ssl? (we've seen it cause problems with ssl, IIRC) 2008-07-13T22:11:03 asjoyner: sorry, that's the basic HTML interface for search (among other things) 2008-07-13T22:11:23 same behavior after fully restarting apache 2008-07-13T22:11:37 miker_: right, the ssl problems that asjoyner is currently encountering are what twigged my memory there 2008-07-13T22:11:55 asjoyner: crap 2008-07-13T22:12:05 we also don't run disk_cache 2008-07-13T22:13:12 oh jiminy cricket 2008-07-13T22:13:23 asjoyner: are you using "start_all" to start up evergreen? 2008-07-13T22:13:37 dbs: yes 2008-07-13T22:13:49 can you paste "ps wax | grep -i open" ? 2008-07-13T22:13:50 dbs: opac login works ... surely cstore is starting 2008-07-13T22:14:35 miker_: true true 2008-07-13T22:14:36 dbs: coming right up 2008-07-13T22:14:49 still a risky thing to have in the docs 2008-07-13T22:15:05