2009-12-21T00:35:18 *** mck9 has left #evergreen 2009-12-21T03:06:08 *** kbsymanz has joined #Evergreen 2009-12-21T06:52:55 *** atz has quit IRC 2009-12-21T07:54:49 *** mck9 has joined #evergreen 2009-12-21T08:03:10 *** artunit has quit IRC 2009-12-21T08:17:27 *** mrpeters-isl has joined #evergreen 2009-12-21T08:36:00 *** berick_ is now known as berick 2009-12-21T08:37:32 *** gdunbar has joined #evergreen 2009-12-21T08:48:07 *** jmeeuwen has quit IRC 2009-12-21T08:49:09 *** finnx2 has quit IRC 2009-12-21T08:49:24 *** finnx has joined #evergreen 2009-12-21T08:56:50 *** bshum has joined #evergreen 2009-12-21T09:18:29 *** phase_bb has joined #evergreen 2009-12-21T09:20:00 *** kbsymanz has quit IRC 2009-12-21T09:26:35 *** atz has joined #evergreen 2009-12-21T09:32:53 *** dchristens has joined #evergreen 2009-12-21T09:33:51 Is there any minimum or maximum character limit for the org unit shortnames? Or is it just a matter of how it appears in the staff client end of things? (i.e. shorter is better?) 2009-12-21T09:34:12 *** jenny has joined #evergreen 2009-12-21T09:35:47 *** dbs has joined #evergreen 2009-12-21T09:43:13 *** bshum has quit IRC 2009-12-21T09:43:27 *** dbs has quit IRC 2009-12-21T09:44:53 *** mck9 has quit IRC 2009-12-21T09:44:53 *** brendan_ga has quit IRC 2009-12-21T09:44:53 *** atheos has quit IRC 2009-12-21T09:44:53 *** phasefx_ has quit IRC 2009-12-21T09:44:53 *** phasefx has quit IRC 2009-12-21T09:44:53 *** phasefx3 has quit IRC 2009-12-21T09:45:47 *** bshum has joined #evergreen 2009-12-21T09:46:37 *** moodaepo has joined #evergreen 2009-12-21T09:47:04 *** dbs has joined #evergreen 2009-12-21T09:49:36 *** mck9 has joined #evergreen 2009-12-21T09:49:36 *** brendan_ga has joined #evergreen 2009-12-21T09:49:36 *** atheos has joined #evergreen 2009-12-21T09:49:36 *** phasefx_ has joined #evergreen 2009-12-21T09:49:36 *** phasefx has joined #evergreen 2009-12-21T09:49:36 *** phasefx3 has joined #evergreen 2009-12-21T10:02:54 *** Godzilla8NJ has joined #evergreen 2009-12-21T10:06:44 *** sfortin has joined #evergreen 2009-12-21T10:25:16 *** bshum_ has joined #evergreen 2009-12-21T10:42:31 *** bshum has quit IRC 2009-12-21T10:42:32 *** bshum_ is now known as bshum 2009-12-21T10:50:01 *** bshum has quit IRC 2009-12-21T10:53:58 *** bshum has joined #evergreen 2009-12-21T11:04:11 *** Godzilla8NJ has quit IRC 2009-12-21T11:09:40 In an effort to get my poor tired brain around OpenSRFy API goodness, I wrote a quick script to generate an html page of registered-api-function to Perl sub mappings. (I know there are better api tools out there - but the excercise is proving useful to me :-) 2009-12-21T11:10:09 Anyone interested in a peek? :-) 2009-12-21T11:11:28 dchristens pasted "simple OpenSRF api quick-reference" at http://paste.lisp.org/display/92427 2009-12-21T11:12:09 Live (temporarily) at http://mintaka.gotdns.org/api.html 2009-12-21T11:16:39 * dbs peeks 2009-12-21T11:17:07 dchristens: do you know about docgen? 2009-12-21T11:17:30 Only a wee bit... which is to say, no. :-) 2009-12-21T11:17:39 http://dev.gapines.org/opac/extras/docgen.xsl?service=open-ils.cat&all=on&offset=0&limit=25 2009-12-21T11:18:10 Ah, very nice. 2009-12-21T11:18:30 Can that drill down to the code? 2009-12-21T11:18:36 docgen.xsl is packaged with opensrf 2009-12-21T11:18:54 No, I don't think so 2009-12-21T11:19:39 * dchristens adds docgen to his bookmarks 2009-12-21T11:19:47 Thanks, dbs 2009-12-21T11:20:10 dchristens: also note that docgen can only generate that stuff for publicly exposed services 2009-12-21T11:20:31 so open-ils.supercat gets no love from there 2009-12-21T11:21:01 2009-12-21T11:23:27 docgen.xsl is just XSL, so you could teach it to at least link from the package to the corresponding Perl module at svn.open-ils.org - would be harder to get direct links to the appropriate line, though, I think 2009-12-21T11:24:58 Heh - I'm *almost* as familiar with XSL as I am with docgen :-) 2009-12-21T11:37:56 dchristens: that api generator is very snazy! 2009-12-21T11:46:48 * dbs laughs at miker_'s commit message 2009-12-21T11:49:29 miker_: thanks :-) 2009-12-21T11:50:51 dchristens: would it be easy for you to add the signature output similar to docgen's? 2009-12-21T11:52:19 "signature output?" Is that api-level, required argument count, etc? 2009-12-21T11:53:32 *** brendan_ga has quit IRC 2009-12-21T11:53:49 yes ... and there's a signature data structure generated when the methods are registerd ... but, if you're parsing the source it's not going to be available ... hrm 2009-12-21T11:55:56 how do you pull the sig data structure out? Is there an api function? (er, meta-function? gak - brain spinning.....) 2009-12-21T11:56:56 :) ... it's part of the introspection api, and a member on the object that implements the method 2009-12-21T11:57:25 so, if you introspect a specific method part of the ouput will be a signature key in the hash 2009-12-21T11:57:37 very cool! 2009-12-21T11:57:52 and it'll have as much (or as little) about param types and such as was provided at registration time 2009-12-21T11:59:14 it should be easy (??) to have the list_api.pl (which is just a very dumb finite state machine) add an introspect request once it's identified a registered function name.... I think.... :-) 2009-12-21T11:59:23 *** r123 has joined #evergreen 2009-12-21T12:00:38 I'll play with that this afternoon :-D 2009-12-21T12:02:38 (wait a sec... wouldn't all of that information be in the __PACKAGE__->register_function stuff? How else would the introspection api learn about it?) 2009-12-21T12:10:01 dchristens: there are several formats available for adding a signature, so it's parsed and compiled from various bits of data 2009-12-21T12:10:10 *** brendan_ga has joined #evergreen 2009-12-21T12:10:46 if you can call introspect on the method, that would be simplest, I think, or you could reuse the signature compiler code inside OpenSRF/Application.pm (IIRC) 2009-12-21T12:13:42 thanks, miker_ ... I'll explore the calling-introspect way :-) 2009-12-21T12:14:16 * dchristens runs, late for a meeting 2009-12-21T12:21:01 *** bshum_ has joined #evergreen 2009-12-21T12:21:46 *** bshum has quit IRC 2009-12-21T12:21:46 *** bshum_ is now known as bshum 2009-12-21T12:30:10 *** bshum has quit IRC 2009-12-21T12:38:15 good god, what was in my head when I was working on serials ingest? 2009-12-21T12:39:37 time to give commit bit to senator and take away mine 2009-12-21T12:40:25 heh 2009-12-21T12:49:14 *** sfortin has quit IRC 2009-12-21T12:57:25 huh, getting some segfaults in opensrf router running opensrf trunk 2009-12-21T12:58:05 memory-constrained environment, but not to the point of OOM, and I haven't seen router segfault before 2009-12-21T12:58:06 dbs: is that the memcached thing 2009-12-21T12:58:07 ? 2009-12-21T12:58:34 atz: what do you mean? 2009-12-21T12:58:50 router runs for a while, then seg faults 2009-12-21T12:59:26 it's not a compilation problem (which is the only libmemcached problem that I'm aware of) 2009-12-21T12:59:56 wonder what else has changed recently then? 2009-12-21T13:00:46 mck9 committed some signal-handling changes very recently 2009-12-21T13:02:02 Yes...and it worked fine when I tested. Not sure what's going on. 2009-12-21T13:02:03 http://svn.open-ils.org/trac/OpenSRF/changeset/1878 most recently 2009-12-21T13:02:51 works okay for a while for me, then segfaults (ubuntu karmic). I'll try upping the memory to take that out of the equation 2009-12-21T13:02:52 atz: can you tell what level is segfaulting? There's a parent, and two children, and two grandchildren. 2009-12-21T13:03:17 i don't think I've seen dbs' problem 2009-12-21T13:03:44 Must be the grandchildren. The parent and children terminate pretty quickly 2009-12-21T13:05:50 dmesg says: opensrf_router[15112]: segfault at a91 ip 001dff1b sp bfa9ddf8 error 4 in libc-2.10.1.so 2009-12-21T13:07:08 as noted, I'll bump the RAM to take that out of the equation at least 2009-12-21T13:07:18 Changes to the signal handling shouldn't make a difference until the signal is received 2009-12-21T13:08:28 The only thing I changed in osrf_router.c, which is where the grandchildren runs until it's shut down, was to add the volatile qualifier to the variable that the signal handler updates. 2009-12-21T13:10:24 I can't conceive of how that would cause a segfault. 2009-12-21T13:13:05 *** zarth has joined #evergreen 2009-12-21T13:13:17 *** zarth has left #evergreen 2009-12-21T13:19:37 The other changes I made over the weekend should affect only the parent process, and by your account that has long since terminated by the time something segfaults. 2009-12-21T13:20:28 I'm inclined to think that something is going wrong somewhere else, like in the low-level messaging code or something. 2009-12-21T13:22:10 dbs: Can you tell what kind of event seems to trigger the segfault? E.g. a particular kind of request from a particular kind of client? 2009-12-21T13:22:20 *** bshum has joined #evergreen 2009-12-21T13:22:58 mck9: no, haven't got to that level of granularity. like I said, first step was to just give the machine more memory 2009-12-21T13:23:28 next step, if I can reproduce it, would be to recompile with the debug flag so that I could get more useful info 2009-12-21T13:25:24 Right. Still, even if more memory avoids the segfault, I'd like to know what's going on. Even in low memory conditions, a segfault always means there's a bug somewhere. 2009-12-21T13:26:04 Let me know when you have more information. 2009-12-21T13:27:41 except when it means physical hardware errors 2009-12-21T13:27:51 True. 2009-12-21T13:28:43 We call malloc() only in a wrapper that traps allocation failures and aborts. However we don't extend the same courtesy to strdup(). Maybe we should. 2009-12-21T13:30:04 i'm speaking more from a systems administrator point-of-view. you should probably ignore me :) 2009-12-21T13:33:55 No, I'd love to be able to blame the hardware. Then it isn't my problem. 2009-12-21T13:37:31 *** jenny has quit IRC 2009-12-21T13:37:55 so error 4 = EINTR I guess (after following the long chain of nested includes to get to /usr/include/asm-generic/errno-base.h 2009-12-21T13:40:14 * dbs tries to reproduce some bad input that he might have been asking router to handle 2009-12-21T13:42:42 should be pretty easy to run ther router in gdb 2009-12-21T13:42:51 *** sunny_ has joined #evergreen 2009-12-21T13:42:53 a/run in/attach with/ 2009-12-21T13:43:10 hmm. can't reproduce now that I've bumped the RAM up 2009-12-21T13:43:40 even with my dodgiest coding :) 2009-12-21T13:44:30 I am trying to run my 1st report. I dont see any sources under Database Source Browser. Any suggestion. I am loggen in as admin. Thanks 2009-12-21T13:49:47 bah. got the segfault again. so it doesn't seem to be directly related to bad input 2009-12-21T13:50:06 sunny_: you have /openils/var/xul/server linked to something that exists? 2009-12-21T13:50:33 err, /openils/var/web/xul/server 2009-12-21T13:51:42 *** dbwells has quit IRC 2009-12-21T13:51:46 and http://localhost/reports/fm_IDL.xml resolves to something that looks like XML? 2009-12-21T13:52:01 *** dbwells has joined #Evergreen 2009-12-21T13:57:40 sboyette: mck9: on the hardware front, this is running under virtualbox, so there's always that. 2009-12-21T14:00:30 dbs: the segfault happened more than a few seconds after launching router, right? 2009-12-21T14:00:53 mck9: yep 2009-12-21T14:01:36 So it's presumably a grandchild that's dying. Did both router processes die, or just one? 2009-12-21T14:01:36 things seem to run for a period of time, some requests run happily, then segfault 2009-12-21T14:01:39 just one 2009-12-21T14:13:03 let me check 2009-12-21T14:13:25 dbs: I further assume that you didn't deliberately send a signal to the router, e.g. from the command line with a kill, or an osrf_ctl.sh -a stop_all (which does a kill internally). It died on its own without obvious intervention on your part? 2009-12-21T14:16:02 *** jenny has joined #evergreen 2009-12-21T14:16:52 Also how do I do a MARC EXPORT of bibs...do I have to run a review file. 2009-12-21T14:20:51 mck9: correct 2009-12-21T14:21:03 Thought so, but I had to ask. 2009-12-21T14:21:07 otherwise I wouldn't have bothered to mention it :) 2009-12-21T14:21:29 Well it shouldn't segfault even if you kill it. But that's moot. 2009-12-21T14:26:28 I don't know if it means anything that the error code was EINTR (assuming that the code presented by the message represented the last value of errno). It means that the process got a signal at some point, but getting a signal doesn't necessarily kill you. 2009-12-21T14:27:29 I'm going to stop worrying about it for now, because I don't know where to go with it without more information. I trust you will keep us informed. 2009-12-21T14:32:27 Or other people will start seeing it once they start running opensrf trunk, and will inform us :) 2009-12-21T14:34:07 bah, just had to work through the "enable scripted windows in Internet Explorer to create a new bookbag on the fly" process, which I assume is a known but not particularly cared-about issue 2009-12-21T14:44:15 * dbs creates https://answers.launchpad.net/evergreen/+faq/866 2009-12-21T14:47:07 Hello folks. Here at Hekman Library we went live with circulation last Friday, so are more or less up and running. It went smoothly enough that in some weird way it was a bit of a let-down :) 2009-12-21T14:47:43 Thanks to everyone for all the help we have received! 2009-12-21T14:50:16 That said... I think we have finally hit our first genuine bug. We are using the in-db circ rules, and when trying to renew for a patron with max overdues, we get a "network" error: Exception: OpenSRF::EX 2009-12-21T14:40:56 OpenSRF::Application /usr/local/share/perl/5.10.0/OpenSRF/Application.pm:202 : Bad event name: \n\n", 2009-12-21T14:50:19 "status":500 2009-12-21T14:50:42 Anyone else hit this before? 2009-12-21T14:51:52 dbwells: sorry, we're using scripts, not in-db circ :( 2009-12-21T14:52:17 same here! 2009-12-21T14:52:26 what is the full network error? 2009-12-21T14:52:37 usualy, there is the error and then a debug, no/ 2009-12-21T14:54:53 mrpeters-isl: yes. I think what I have sent is the meat of it, though. 2009-12-21T14:55:25 *** dchristens has left #evergreen 2009-12-21T14:55:43 dbwells: there are fixes in rel_1_6 to prevent that error in the circ code. it would be pretty easy to patch by hand if anyone wants to try.. 2009-12-21T14:55:47 *** sfortin has joined #evergreen 2009-12-21T14:55:48 @DBS...I do have a dir called /openils/var/web/xul/server and it's link target is "rel_1_6_0_0 folder which contains folder like admin, auth..OPENILS....index.xhtml 2009-12-21T14:55:48 sunny_: Error: No closing quotation 2009-12-21T14:56:04 Sorry..I had to go to the server room to look for the directory 2009-12-21T14:56:33 berick: sure, I am up for it. Any particular patch set? 2009-12-21T14:56:35 basically, every time the code accesses '$LEGACY_CIRC_EVENT_MAP->{$_->{fail_part}}' make sure it's followed by ' || $_->{fail_part}' 2009-12-21T14:56:42 in Circulate.pm 2009-12-21T14:57:01 IOW: 2009-12-21T14:57:02 $LEGACY_CIRC_EVENT_MAP->{$_->{fail_part}} || $_->{fail_part}) 2009-12-21T14:57:16 there are a few cases in rel_1_6_0 where that is not the case 2009-12-21T14:57:51 dbwells: i think the fix got in with other changes, iow no patchset that won't potentially change other stuff 2009-12-21T14:59:04 looks like just 2 instances.. around lines 910 and 2420 2009-12-21T15:00:21 berick: will try and report back, thanks! 2009-12-21T15:00:35 k 2009-12-21T15:01:47 *** sfortin has quit IRC 2009-12-21T15:09:04 berick: well, that did it. Thanks again. I do get some odd behavior when trying to do renew more than one at a time, but not an error, so I can live with that. 2009-12-21T15:09:47 dbwells: great, thanks for trying it 2009-12-21T15:10:20 * berick will make sure that's part of the 1.6.0.1 release 2009-12-21T15:18:22 berick++ dbwells++ 2009-12-21T15:24:37 *** brendan_ga has quit IRC 2009-12-21T15:31:48 Well, here's another oddity. I have renewed all of this user's items, but he is still marked as having Maximum Overdues. In the Standing box, "Patron exceeds max overdue item threshold" is listed three times, but the Overdue count lists '0'. Ideas? 2009-12-21T15:35:38 "Refresh" in the patron view should in theory re-calc those... I think. 2009-12-21T15:35:49 Odd about the triplication, though. 2009-12-21T15:42:55 I addition to running a EG Postgres slave server for reports any suggestions on how to optimize the file system that houses the PostgreSQL database? Is it worth it? 2009-12-21T15:43:28 8k blocks, not using journaling, setting the fs to noatime (?), thoughts, anyone? 2009-12-21T15:54:01 *** bshum has quit IRC 2009-12-21T16:01:53 *** pmplett has joined #evergreen 2009-12-21T16:04:03 Well, I have tried several things and no luck, he is frozen somehow with Maximum Overdues. Very odd... 2009-12-21T16:08:48 Ok, I got around it for now by going to Messages and selecting "Remove from patron" for each penalty. Probably just a newbie mistake, but in two other test cases the penalty went away on its own when renewing. Oh well, we'll see if it recurs. 2009-12-21T16:18:32 perhaps what would normally have "cleared" the penalty was being confused that there were three for the same reason. doesn't explain how the three got there in the first place, of course. 2009-12-21T16:25:47 Has anyone looked into permitting patrons to self-identify their "home" library? 2009-12-21T16:26:03 We'd want a hard boundary on how far they could jump in the org tree... 2009-12-21T16:46:31 *** pmplett is now known as pmp_afk 2009-12-21T16:49:55 *** brendan_ga has joined #evergreen 2009-12-21T16:53:27 *** dchristens has joined #evergreen 2009-12-21T16:55:53 *** finnx has quit IRC 2009-12-21T17:08:17 *** r123 has left #evergreen 2009-12-21T17:26:26 *** dchristens has left #evergreen 2009-12-21T17:28:34 *** sunny_ has quit IRC 2009-12-21T17:30:46 *** jenny has left #evergreen 2009-12-21T18:09:22 *** dbs has quit IRC 2009-12-21T18:56:54 *** pmp_afk is now known as pmplett 2009-12-21T21:05:20 *** wlayton has joined #evergreen 2009-12-21T21:52:45 *** finnx has joined #evergreen 2009-12-21T22:03:12 *** jamesrf has joined #evergreen 2009-12-21T22:21:22 *** wlayton has quit IRC 2009-12-21T23:04:51 *** dbs has joined #evergreen