| Time |
Nick |
Message |
| 00:54 |
|
sandbergja joined #evergreen |
| 06:00 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 07:18 |
|
rjackson_isl_hom joined #evergreen |
| 08:08 |
|
mantis joined #evergreen |
| 08:14 |
|
collum joined #evergreen |
| 08:39 |
|
mmorgan joined #evergreen |
| 08:41 |
|
collum_ joined #evergreen |
| 08:48 |
|
mmorgan joined #evergreen |
| 08:54 |
|
Dyrcona joined #evergreen |
| 09:15 |
|
dbwells joined #evergreen |
| 09:18 |
|
jvwoolf joined #evergreen |
| 09:23 |
|
alynn26 joined #evergreen |
| 09:25 |
Dyrcona |
ldirectord-- |
| 09:25 |
Dyrcona |
Is anyone successfully using HAProxy instead? I tried it once, but the performance was awful. |
| 09:26 |
Dyrcona |
I ask because we had a SIP server go down on Saturday, and ldirectord kept trying to send it requests. I'm looking into have ldirectord monitor and remove them automatically. |
| 09:27 |
Dyrcona |
When I tested haproxy on a large set of vms, I had written a custom monitor program that would let haproxy remove a SIP server when it was not responding. |
| 09:32 |
Dyrcona |
Bleh. Forty-two not connected to the network errors from 1 brick drone. Is there a way to determine the service or client from those messages? It's almost always something like this: opensrf bh6.public/_bh6_1620048251.912167_5029 |
| 09:32 |
Dyrcona |
Also, this seems to have gotten worse since we upgraded to 3.5.3. Anyone else noticed this? |
| 10:08 |
Dyrcona |
Well, it looks like I can repurpose the program that I wrote for haproxy to monitor SIP2 for ldirectord with only a very minor change to the command line argument parsing. |
| 10:24 |
jvwoolf |
Dyrcona: We tried HAProxy once and it didn't work for us either. As far as I know, we haven't had any ldirectord issues since our upgrade to 3.5.4. I think we only use it on on our app servers though, not our SIP servers. |
| 10:25 |
Dyrcona |
jvwoolf: Thanks. |
| 10:25 |
Dyrcona |
Also, I don't have to change the program at all. I thought I did, but it didn't work. I went back to the original and it works. :) |
| 10:26 |
Dyrcona |
I suspect haproxy would work better if it didn't do ssl to the apache servers. |
| 10:26 |
Dyrcona |
ldirectord is just passing the requests along whereas haproxy is doing SSL both ways. |
| 10:34 |
Dyrcona |
Since I am messing with the ldirectord config, I wonder if I want to switch to quiescent=no. quiescent=yes seems all right if we're deliberately dropping a service and will wait before taking it completely offline, but not for crashes. |
| 10:57 |
Dyrcona |
Grr... cstore, reporter-store, and pcrud don't connect to my test database server this morning and the configuration is correct, AFAICT. |
| 10:59 |
Dyrcona |
Oh, just brilliant: May 3 10:56:53 jasontest open-ils.cstore: [INFO:7279:oils_sql.c:210:] open-ils.cstore successfully connected to the database |
| 10:59 |
Dyrcona |
but --dianostic says it has a pid file but no processes running. |
| 11:02 |
Dyrcona |
Too slooowww. |
| 11:04 |
jeff |
does the pid file match the pid of the process? did you run into the "fork fails and listener becomes a drone" issue? |
| 11:08 |
|
jihpringle joined #evergreen |
| 11:15 |
Dyrcona |
jeff: Don't think so. I tried it a few times with start-all and stop-all. Eventually, I just waited long enough... :) |
| 11:16 |
Dyrcona |
I was starting services to see if SIPServer responds to status messages without the client logging in first. Doesn't look like it does. |
| 11:19 |
|
jvwoolf1 joined #evergreen |
| 11:22 |
Dyrcona |
That would be handy for something like I'm trying to do. As it is, my check program has to login to sipserver and that leaves a bunch of orphaned auth sessions. |
| 11:30 |
alynn26 |
ll |
| 12:15 |
|
Christineb joined #evergreen |
| 13:14 |
|
jvwoolf joined #evergreen |
| 14:07 |
JBoyer_ |
Dyrcona, You might try SC Status, message 99. Since it is sometimes sent before Login (93) SIPServer tries to respond to it even if you're not logged in. (It grabs the first specified login/institution to do something. |
| 14:08 |
|
jihpringle95 joined #evergreen |
| 14:08 |
Dyrcona |
JBoyer_: That's what I was trying, but I realize now, that I busted the message. |
| 14:08 |
* Dyrcona |
tries again. |
| 14:09 |
JBoyer_ |
Ah, I see what you mean now. I must have skimmed that line. |
| 14:10 |
JBoyer |
It's supposed to anyway, who knows if the comments and the code disagree. :) |
| 14:11 |
Dyrcona |
I'm sending 9902552.00\r (where \4 is carriage return), and I get no response from the server. |
| 14:12 |
Dyrcona |
\r is CR... |
| 14:14 |
|
dbwells joined #evergreen |
| 14:17 |
Dyrcona |
My program to login works, though. |
| 14:33 |
Dyrcona |
Jamming a | before the \r doesn't help, either. |
| 14:34 |
Dyrcona |
S'pose I could try adding a sequence and checksum, but login works without them. |
| 14:58 |
Dyrcona |
Nope. Guess you have to login, first. |
| 15:00 |
Dyrcona |
Hm... The login program isn't working on the vm, now. |
| 15:08 |
Dyrcona |
That's bizarre. It worked when I statically linked the binary. |
| 15:08 |
Dyrcona |
The login program that is. |
| 15:09 |
Dyrcona |
The one that sends the SC status message still doesn't work. :( |
| 15:10 |
Dyrcona |
I should add that I'm building and running these on my laptop. |
| 15:10 |
Dyrcona |
Not on the vm.... |
| 15:14 |
Dyrcona |
Yeah, that's bizarre. I swear a dynamically linked binary worked this morning, but now, it's only working if statically linked, on the machine where I compiled it. |
| 15:15 |
Dyrcona |
The code hasn't changed either. I just added a couple of comments. |
| 15:17 |
Dyrcona |
"Curiouser and curiouser," said Alice. -- So a dynamically linked binary works if I specify -O2 optimization. |
| 15:20 |
Dyrcona |
gcc bugs? |
| 15:21 |
Dyrcona |
@monologue |
| 15:21 |
pinesol |
Dyrcona: Your current monologue is at least 17 lines long. |
| 15:22 |
|
dbwells_ joined #evergreen |
| 15:50 |
|
mantis left #evergreen |
| 15:55 |
|
jeff_ joined #evergreen |
| 15:55 |
|
devted_ joined #evergreen |
| 15:55 |
|
sandbergja joined #evergreen |
| 16:06 |
Bmagic |
what causes an EDI invoice to get assigned to a branch as the receiver? When the PO is at the system level. The EDI account is at the system level. Everything that I can find is at the system level. But the EDI generated Invoice was assigned to a seemly random branch inside the system |
| 16:08 |
Bmagic |
I might have found it. The SAN lookup on actor.org_address matching this branch perhaps? |
| 16:09 |
Dyrcona |
Most likely |
| 16:11 |
Bmagic |
Drycona++ # always johnny on the spot! |
| 16:12 |
Dyrcona |
I don't know, but it that's what you found, then it probably is the SAN. I know that's used if it is available. |
| 16:12 |
Bmagic |
EDI.pm led me to that answer |
| 16:12 |
Bmagic |
sub process_message_buyer { |
| 16:12 |
Dyrcona |
Yeah. |
| 16:33 |
|
dbwells joined #evergreen |
| 16:55 |
|
jvwoolf joined #evergreen |
| 17:07 |
|
dbwells joined #evergreen |
| 17:27 |
|
mmorgan left #evergreen |
| 17:38 |
|
dbwells joined #evergreen |
| 18:00 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 18:07 |
|
Dyrcona joined #evergreen |
| 18:08 |
Dyrcona |
@later tell JBoyer I got the status to work after logging in, first, so there was nothing wrong with my status message. It's just not working with Evergreen unless one logs in. |
| 18:08 |
pinesol |
Dyrcona: The operation succeeded. |
| 18:09 |
JBoyer |
Poof. |
| 18:09 |
Dyrcona |
Heh. I was just about to sign out of IRC, again. :) |
| 18:10 |
JBoyer |
That's annoying, I was able to use it against one of our hosted systems without any trouble, but my dev server doesn't have functioning SIP so I wasn't able to get it working there. |
| 18:10 |
Dyrcona |
Anyway, since just logging in is enough to check if a SIPServer is functioning OK, I'm going to stick with the first program that I wrote about 2 years ago. |
| 18:11 |
JBoyer |
But yeah, so long as the logging in isn't too frequent it shouldn't tax anything. |
| 18:11 |
Dyrcona |
There might be an option? |
| 18:11 |
JBoyer |
I didn't have time to look into it that much today. |
| 18:11 |
Dyrcona |
One every 5 mintes to start. |
| 18:12 |
JBoyer |
I didn't mean to stretch out your day though, so long as you have something that works that's good. |
| 18:12 |
JBoyer |
Dyrcona++ |
| 18:12 |
Dyrcona |
I think I'll look some more, now, even though I'm off the clock. I felt like messing with the C code some more. I'd write a library for 3M SIP2 in C, but the world probably doesn't need another 3M SIP2 library. (I came across one in Java today.) |
| 18:13 |
Dyrcona |
Well, it's more from my own personal curiosity at this point. |
| 18:13 |
JBoyer |
At this point I expect SIP2 to die when MARC itself does, so how much the world needs another lib depends on how much you dislike the other options, heh. |
| 18:14 |
Dyrcona |
Yeah. |
| 18:15 |
Dyrcona |
I'm OK with PHPSIP2 for most things. I wrote this C program to be used by haproxy, and it turns out to work with liderectord, too. |
| 18:24 |
Dyrcona |
Hmm. Looking at the code, I may have a problem with the configuration file. |
| 18:24 |
Dyrcona |
Looks like SIPServer will try to load the configuration for the "first" account if you send sc status without logging in. |
| 18:31 |
|
sandbergja joined #evergreen |
| 18:31 |
Dyrcona |
Except, I can't find any of the error messages in the logs on the test vm. |
| 18:38 |
Dyrcona |
May 3 11:11:29 jasontest opensrf_sip[9672]: raw_transport: LOGIN ERROR: 'raw_transport: sending SC status before login not enabled, exiting at SIPServer.pm line 578.#012' |
| 18:39 |
Dyrcona |
It is an option. Guess I'll find it and enable it. |
| 18:42 |
|
jonadab joined #evergreen |
| 18:51 |
Dyrcona |
"And there was much rejoicing!" |
| 18:52 |
Dyrcona |
JBoyer++ jeff++ |
| 19:07 |
|
jvwoolf joined #evergreen |
| 19:08 |
Dyrcona |
Now that I have something that I can set up that won't eat up auth sessions, my day is truly done. Catch you all again tomorrow! |
| 19:16 |
|
jvwoolf1 joined #evergreen |
| 20:05 |
|
jvwoolf1 left #evergreen |
| 23:09 |
|
jvwoolf joined #evergreen |
