| Time |
Nick |
Message |
| 05:02 |
pinesol_green |
News from qatests: Test Success <http://testing.evergreen-ils.org/~live> |
| 06:47 |
|
tspindler joined #evergreen |
| 06:57 |
|
agoben joined #evergreen |
| 07:13 |
|
rjackson_isl joined #evergreen |
| 08:11 |
|
collum joined #evergreen |
| 08:24 |
|
mdriscoll joined #evergreen |
| 08:38 |
|
mmorgan joined #evergreen |
| 09:06 |
csharp |
I'm facing the apparmor issue with ejabberd on xenial that came up a few months ago and I can't remember the solution :-/ |
| 09:07 |
|
Dyrcona joined #evergreen |
| 09:13 |
|
kmlussier joined #evergreen |
| 09:18 |
|
Dyrcona joined #evergreen |
| 09:21 |
csharp |
maybe it's not apparmor - I set ejabberdctl into "complain mode" but it's still not letting opensrf start |
| 09:24 |
JBoyer |
csharp, no complaints in any opensrf logs? |
| 09:24 |
|
yboston joined #evergreen |
| 09:25 |
csharp |
Use of uninitialized value $@ in concatenation (.) or string at /usr/local/share/perl/5.22.1/OpenSRF/Transport/SlimJabber/Client.pm line 162. |
| 09:25 |
csharp |
Exception: OpenSRF::EX::Jabber 2017-01-24T09:25:07 OpenSRF::Transport::SlimJabber::Client /usr/local/share/perl/5.22.1/OpenSRF/Transport/SlimJabber/Client.pm:162 Jabber Exception: Could not authenticate with Jabber server: |
| 09:25 |
csharp |
is all I have to go on |
| 09:25 |
JBoyer |
:/ |
| 09:26 |
csharp |
/openils/bin/osrf_control[1611]: [ERR :1611:XMPPReader.pm:131:] XMPP connect failed |
| 09:27 |
csharp |
Jan 24 09:24:32 governor-xenial gateway[1459]: [WARN:1459:socket_bundle.c:369:] socket_open_tcp_client(): Cannot connect to server public.localhost: Connection refused |
| 09:27 |
csharp |
that's what led me to think it was apparmor |
| 09:27 |
csharp |
ufw isn't running |
| 09:29 |
csharp |
interesting that the connection refused messages are all about public.localhost - none for private.localhost, but maybe it need to connect to public before trying private? |
| 09:30 |
csharp |
ejabberd is running and ejabberdctl commands work |
| 09:31 |
|
mmorgan joined #evergreen |
| 09:32 |
csharp |
hmm Failed legacy authentication for opensrf private.localhost/client_at_localhost_1611 |
| 09:35 |
JBoyer |
Yeah, I don't think $@ is the right magic thing to catch a "your username or password are wrong" kind of error there. The description is vauge. |
| 09:36 |
csharp |
I just unregistered/re-registered my ejabberd users |
| 09:36 |
Dyrcona |
csharp: Is public.localhost typoed in the hosts or the ejabberd config? |
| 09:37 |
csharp |
no, they look right to me |
| 09:38 |
csharp |
ah |
| 09:38 |
csharp |
Jan 24 09:36:01 governor-xenial kernel: audit: type=1400 audit(1485268561.667:32): apparmor="DENIED" operation="connect" profile="/usr/sbin/ejabberdctl//su" name="/run/dbus/system_bus_socket" pid=2360 comm="su" |
| 09:38 |
csharp |
so it is apparmor |
| 09:39 |
csharp |
/usr/sbin/ejabberdctl//su is still in enforce mode |
| 09:39 |
csharp |
I bet changing that does it |
| 09:39 |
Dyrcona |
Most likely. |
| 09:39 |
Dyrcona |
I've not run into that. |
| 09:40 |
|
krvmga joined #evergreen |
| 09:41 |
Dyrcona |
Well, I thought I was going to have an easy morning on enable apache on a server and point z39.50 at it. Turns out, I have to configure the apache, too. :) |
| 09:47 |
JBoyer |
Darn computers, they only do what I say, not what I want. :/ |
| 09:48 |
Dyrcona |
:) |
| 09:48 |
Dyrcona |
Well, I thought it had already been configured here, but I was wrong. |
| 09:48 |
csharp |
@quote add < JBoyer> Darn computers, they only do what I say, not what I want. :/ |
| 09:48 |
pinesol_green |
csharp: The operation succeeded. Quote #162 added. |
| 09:50 |
Dyrcona |
Permissions on the ssl key is a bit annoying....Makes copying by scp a bit trickier, but I guess that's the point. :) |
| 09:53 |
|
jlundgren joined #evergreen |
| 09:56 |
Dyrcona |
I think oils_ctl.sh should be modified so it can start z39.50 as root. IIRC, it bails if it isn't run as the opensrf user. |
| 09:57 |
Dyrcona |
I think it and/or opensrf-perl.pl need some other changes so they're less fragile when run. I know they also fail if $PREFIX/bin is not in the PATH. |
| 09:58 |
Dyrcona |
I've meant to LP those ideas for a while, now, but no time. |
| 09:58 |
|
mmorgan1 joined #evergreen |
| 09:58 |
* JBoyer |
feels for Dyrcona. Many are the LPs in my head, but faster are the workarounds and more work waits. :( |
| 10:22 |
|
Christineb joined #evergreen |
| 10:28 |
Dyrcona |
Great....The two servers where I've tried yaz-client get connection refused, but someone is connecting because the number of simple2zooms > 1. |
| 10:28 |
Dyrcona |
I just wanted to test the configuration, but I don't remember from where I tested it successfully last time. |
| 10:29 |
Dyrcona |
Maybe the training server? |
| 10:52 |
|
jvwoolf joined #evergreen |
| 11:23 |
|
abowling joined #evergreen |
| 11:27 |
abowling |
question: are there guidelines anywhere published for PO completion in order to get it committed into master? |
| 11:28 |
abowling |
i ask because we're working on getting multiple languages rolled in and it would very useful to have. my question is whether the community has established % completion, etc. |
| 11:37 |
|
khuckins joined #evergreen |
| 11:38 |
Dyrcona |
On a brick setup, should the drones really mount /openils from the brick head? |
| 11:39 |
Dyrcona |
Or, rather, do OpenSRF drones need any of the files generated by autogen? |
| 11:40 |
|
_adb joined #evergreen |
| 11:48 |
|
dcz_ joined #evergreen |
| 11:48 |
Dyrcona |
abowling: I'm not aware of any published guidelines. We take translations as they come, though we did talk about deleting some below a certain % complete. I don't know what came of that. |
| 11:49 |
abowling |
dyrcona: thanks. |
| 11:50 |
abowling |
we'd be real interested in knowing before any were deleted, because we've got plans to flesh out quite a few in coming months |
| 11:52 |
|
mmorgan joined #evergreen |
| 11:54 |
Dyrcona |
bshum and dbs seem to have the most to do with translations lately, but we don't really have an i18n team AFAIK. It's in the beginning stages maybe. |
| 11:57 |
miker |
Dyrcona: re "run stuff as user other than opensrf" we maintain a patch for exactly that, and also a "tell autogen where the config file is, instead of assuming a location" patch. I'm pretty sure I've posted the former, but I'll check |
| 11:57 |
JBoyer |
Yay initial auth work, 50MB scripts... D: |
| 11:58 |
Dyrcona |
miker: We've had some weirdness going on with copies being "deleted" since we put a new brick in place yesterday. I'm starting to suspect it is because the drones weren't sharing /openils from the brick head. |
| 11:59 |
Dyrcona |
I've added the nfs share/mounts and I'll see if it settles down. |
| 12:00 |
Dyrcona |
My logs seems to be spotty, still, but I found a copy update of one copy from the time that the db basically says the copy was deleted yesterday. I think the IDL is out of whack on the drones, because auogen.sh was run on the brick head. |
| 12:00 |
Dyrcona |
Anyway.... I'm hoping the nfs share fixes it. |
| 12:05 |
|
brahmina joined #evergreen |
| 12:11 |
Dyrcona |
And, no. The new brick's IDL is not the same as the others. This is what happens when you make changes that don't get added to the proper git repos. |
| 12:13 |
Dyrcona |
Now, I'm back to being not sure about needing that nfs mount... :) |
| 12:19 |
Dyrcona |
Oh, nice. And the fm_IDL.xml in /openils/var/web/reports is yet another size. |
| 12:21 |
dbs |
entityized fm_IDL.xml perhaps? |
| 12:21 |
dbs |
we tried to get an i18n coordinator back at the vancouver conference :/ |
| 12:25 |
Dyrcona |
dbs: yeah, that's probably it, but I put the other one in place anyway. |
| 12:25 |
Dyrcona |
I can see splitting the IDL up into chunks making this harder to mange. :( |
| 12:29 |
dbs |
I blame XML |
| 12:31 |
dbs |
would be nice to teach unapi to return JSON, including MARC-in-JSON, to avoid XML parsing etc on clients - heh |
| 12:31 |
Dyrcona |
Bleah... Branch overload! |
| 12:33 |
kmlussier |
dbs: I think we have an i18n coordinator now. Eva. |
| 12:36 |
dbs |
oh?! yay |
| 12:36 |
dbs |
<-- too tuned out |
| 12:38 |
|
bmills joined #evergreen |
| 12:42 |
JBoyer |
csharp, I meant to ask earlier, did you get that apparmor error message from the journal, or is 16.04 still using text based syslog? |
| 12:50 |
|
sandbergja joined #evergreen |
| 13:01 |
|
jihpringle joined #evergreen |
| 13:03 |
bshum |
JBoyer: I asked csharp privately but wondering to verify whether the auth type was set to plain for legacy connection in ejabberd config, since we're talking xenial. |
| 13:04 |
bshum |
Otherwise, the hash it does by default won't work with opensrf and it dies an unhappy death with unable to connect |
| 13:04 |
bshum |
Similar to the errors being described earlier in the log. |
| 13:05 |
bshum |
"auth_password_format" to plain ; from the ejabberd 16.x instruction for Ubuntu Xenial. http://evergreen-ils.org/documentation/install/OpenSRF/README_2_5_0_alpha.html#_configure_the_ejabberd_server |
| 13:08 |
bshum |
abowling: Dyrcona: dbs: for i18n, there was this email thread awhile back where we outlined some ideas on how to do translations stuff going forward: http://georgialibraries.markmail.org/thread/sthmnbvfpt47pe7n |
| 13:08 |
bshum |
But that's still work in progress too. Probably should talk to Eva about getting more formal procedures into place for language support in Evergreen. |
| 13:09 |
bshum |
It's pretty wild west. I usually do whatever the localization lead tells me to do :) |
| 13:09 |
bshum |
Example: You say Spanish is done? Okay, awesome! |
| 13:14 |
* bshum |
wanders off |
| 13:38 |
csharp |
JBoyer: journal |
| 13:39 |
csharp |
also, looks like apparmor isn't the problem - now looking over bshum's comments |
| 13:40 |
|
remingtron joined #evergreen |
| 13:41 |
csharp |
I was not using the right password format, but changing it hasn't solved the problem yet |
| 13:42 |
csharp |
I removed apparmor via APT to rule it out |
| 13:43 |
csharp |
ok - it was the password format |
| 13:46 |
csharp |
since I was installing OpenSRF 2.4.1, I wasn't looking at the right instructions |
| 13:51 |
|
collum_ joined #evergreen |
| 14:07 |
Dyrcona |
Woot! I got the ssh tunnel feature of PgAdmin to actually work! |
| 14:07 |
Dyrcona |
I used to make tunnels of my own and connect to those. |
| 14:45 |
|
pgardella joined #evergreen |
| 14:48 |
Dyrcona |
What a day... |
| 14:48 |
Dyrcona |
So, I installed the prereqs for NCIPServer on the model VMs, but then forgot to install NCIPServer.... |
| 14:48 |
Dyrcona |
Just copying it from another brick works, though. |
| 14:48 |
Dyrcona |
Nothing like putting it in production to find out what you missed. :) |
| 14:51 |
kmlussier |
Wheee! |
| 14:51 |
Dyrcona |
Now, what to do about 4,882 copies that were updated on the brick head with the busted IDL.... |
| 14:53 |
jeff |
i see some queries involving auditor.asset_copy_history in your future. |
| 14:56 |
|
collum joined #evergreen |
| 14:59 |
Dyrcona |
Hmm... Can I tell psql to fill a variable from a file, one line at a time? Something tells me, "No." |
| 15:00 |
berick |
recently vacuum-full'ed auditor.asset_copy_history on a test server after removing old entries... table size went from 127G to 11G. woot. |
| 15:00 |
berick |
in less than 20 minutes, no less |
| 15:02 |
|
mmorgan1 joined #evergreen |
| 15:03 |
csharp |
@love vacuum full |
| 15:03 |
pinesol_green |
csharp: The operation succeeded. csharp loves vacuum full. |
| 15:04 |
pgardella |
Afternoon everyone! I was looking at our Evergreen DB, and saw that the real_full_rec table is taking up 32GB (according to Postgres) (31,390,016 records). Should it be that big? |
| 15:05 |
berick |
pgardella: 70G here |
| 15:05 |
berick |
4th largest table |
| 15:06 |
berick |
mileage may vary |
| 15:06 |
pgardella |
@berick It's our largest table by far. The next biggest is combined_keyword_field_entry at 4 GB |
| 15:06 |
pinesol_green |
pgardella: As great as you are man, you'll never be greater than yourself. |
| 15:07 |
pgardella |
Just wanted to make sure nothing was wrong, like we're not cleaning something up |
| 15:08 |
berick |
no, it's just big |
| 15:08 |
Dyrcona |
phasefx++ for mentioning that O'Reilly sysadmin humble bundle. It has paid for itself. |
| 15:09 |
Dyrcona |
pgardella: Yes, it's big. There's a row for ever subfield of every MARC record. |
| 15:10 |
phasefx |
Dyrcona: awesome |
| 15:10 |
Dyrcona |
s/ever/every/ |
| 15:10 |
tsbere |
Plus the leader and some other data ;) |
| 15:10 |
Dyrcona |
Right, that, too. |
| 15:12 |
pgardella |
works for me. |
| 15:13 |
Dyrcona |
When I restore dumps, the very last thing to finish (hours after everything else) is the index on metabib.real_full_rec.index_vector. |
| 15:41 |
|
mmorgan joined #evergreen |
| 16:46 |
Stompro |
Hello, a staff member just had a problem pulling up a patron account via the staff client patron search. The customer is a business and has the name "420 insurance services". But it doesn't seem possible to search for a first or last name that starts with a number. Is that behavior configurable? |
| 16:47 |
|
jlundgren left #evergreen |
| 16:54 |
Dyrcona |
And, now, I remember why I used my own ssh tunnels with PgAdmin. When the ssh connection is silently dropped behind your back, PgAdmin crashes. |
| 16:54 |
Dyrcona |
It apparently has no concept of keep alive packets. |
| 16:55 |
phasefx |
Stompro: it's the staff client doing that; the web staff client isn't so zealous. Look at patron/search_form.js, line 372 |
| 16:57 |
phasefx |
Stompro: you could comment that out on the server-side and be okay, I think |
| 17:02 |
pinesol_green |
News from qatests: Test Success <http://testing.evergreen-ils.org/~live> |
| 17:08 |
|
mmorgan left #evergreen |
| 17:12 |
Stompro |
Thanks, I'm trying to figure out why this is done. it would be one thing if names starting/ending with numbers couldn't be registered... |
| 17:19 |
Stompro |
Pre commit f7db7f578e6f numbers were allowed, then it was changed to allow unicode characters but to exclude digits. I wonder if that was on purpose or not? |
| 17:19 |
pinesol_green |
Stompro: [evergreen|dbs] Use Unicode-friendly regular expressions for registering and retrieving patrons by name - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=f7db7f5> |
| 17:20 |
phasefx |
it looks like it originally stripped only leading whitespace |
| 17:21 |
Stompro |
and punctuation I think. |
| 17:22 |
phasefx |
hrmm.. if not a "word character" or space, strip. Poor commit message on my part; don't remember who lobbied for it |
| 17:24 |
phasefx |
or why |
| 17:25 |
Stompro |
I think non word character matches [^a-zA-Z0-9_] so underscores were allowed also. |
| 17:27 |
Stompro |
I'll send a message to the list to see if anyone thinks that digits should still be stripped. |
| 17:28 |
Stompro |
phasefx++ thanks for the help |
| 17:28 |
phasefx |
sorry I couldn't help more, but you're welcome |
| 17:28 |
phasefx |
looks like we were using bugzilla back then, and I think all that data is gone gone :) |
| 17:46 |
Bmagic |
Is there a library setting that prevents the hold targeter from choosing another copy for a certain amount of time (default 24 hours) |
| 17:49 |
Bmagic |
Better yet, is there a mechanism to keep the targeted copy through hours of operation (closed). We have a scenario where the hold would be filled faster if it would continue to target a copy at the closer branch through it's closed hours |
| 17:52 |
berick |
Bmagic: you can change the retarget interval by modifying the hold_targeter.pl. it's not per-library though |
| 17:52 |
Bmagic |
I see |
| 17:52 |
Bmagic |
Thank you! |
| 17:55 |
|
dcook joined #evergreen |
| 18:10 |
|
brahmina joined #evergreen |
| 19:54 |
|
jvwoolf joined #evergreen |
| 21:16 |
|
jvwoolf joined #evergreen |
