| Time |
Nick |
Message |
| 05:00 |
pinesol_green |
News from qatests: Test Failure <http://testing.evergreen-ils.org/~live> |
| 07:16 |
|
rjackson_isl joined #evergreen |
| 07:21 |
|
JBoyer joined #evergreen |
| 08:06 |
|
collum joined #evergreen |
| 08:41 |
|
mmorgan joined #evergreen |
| 08:48 |
|
kmlussier joined #evergreen |
| 09:01 |
|
yboston joined #evergreen |
| 09:02 |
|
bos20k joined #evergreen |
| 09:21 |
|
jvwoolf joined #evergreen |
| 09:48 |
|
mmorgan1 joined #evergreen |
| 09:56 |
|
sarabee joined #evergreen |
| 10:10 |
|
maryj_ joined #evergreen |
| 10:10 |
|
maryj joined #evergreen |
| 10:11 |
|
maryj joined #evergreen |
| 10:37 |
|
Christineb joined #evergreen |
| 11:04 |
|
mmorgan joined #evergreen |
| 11:10 |
|
DPearl joined #evergreen |
| 11:21 |
|
bmills joined #evergreen |
| 12:08 |
|
maryj joined #evergreen |
| 12:12 |
|
sandbergja joined #evergreen |
| 12:17 |
|
brahmina joined #evergreen |
| 12:32 |
|
mmorgan1 joined #evergreen |
| 12:33 |
* tsbere |
dislikes certain kinds of errors |
| 12:33 |
tsbere |
Like this one: ERROR: index row size 2824 exceeds maximum 2712 for index "metabib_full_rec_value_idx" |
| 12:34 |
* tsbere |
assumes that is due to a very long 520a in the record in question |
| 12:47 |
jeffdavis |
I've run into that kind of thing before |
| 12:47 |
|
kmlussier joined #evergreen |
| 12:48 |
|
barbara_ joined #evergreen |
| 12:54 |
tsbere |
jeffdavis: In this case it appears to be a fairly long summary in Chinese. <_< |
| 12:58 |
jeffdavis |
yeah I don't have the specifics handy but we've had to modify at least one such index to base it on a substring of the full value |
| 13:28 |
dbs |
tsbere: ahh. I wonder if its a bytes-vs-char thing, given that the stock definition is "metabib_full_rec_value_idx" btree ("substring"(value, 1, 1024)) |
| 13:30 |
dbs |
or just how a btree index deals with multibyte characters (seems like poorly) |
| 13:31 |
tsbere |
dbs: Given that this summary is almost all unicode.....I have the entire thing in wordpad copied from flat-text mode if you want me to drop it somewhere. |
| 13:44 |
|
jihpringle joined #evergreen |
| 14:19 |
* jeff |
waits for Slack to swap in |
| 14:19 |
jeff |
oof. |
| 14:20 |
jeff |
good morning! |
| 14:20 |
tsbere |
jeff: A bit late for a good morning from my POV ;) |
| 14:20 |
jeff |
true. |
| 14:20 |
jeff |
good afternoon! |
| 14:23 |
* kmlussier |
waves to jeff |
| 14:26 |
|
bmills joined #evergreen |
| 14:31 |
|
Dyrcona joined #evergreen |
| 14:32 |
jeff |
I do need to look into switching NCIP temporary items to pre-cats. |
| 14:32 |
jeff |
i grow weary of excluding the bibs in question from things like QA reports :-) |
| 14:42 |
Dyrcona |
So, it appears that sending a call number via SMS from the catalog is not working for us. |
| 14:42 |
Dyrcona |
However, the event shows as complete and the event output looks good. |
| 14:43 |
Dyrcona |
I get no log entry in my mail server for the message being sent. |
| 14:43 |
Dyrcona |
SMS hold notifications do get a log entry and say they're complete and the output looks good. |
| 14:44 |
Dyrcona |
Anyone got any ideas? |
| 14:48 |
Dyrcona |
jeff: I'm not sure pre-cats are really ready for that use case. Last time I tried, I had "issues." Unfortunately, I am not able to elaborate at this time. |
| 14:54 |
|
jvwoolf left #evergreen |
| 15:00 |
tsbere |
Dyrcona: Did you get a from address in the output? |
| 15:03 |
Dyrcona |
tsbere: Yes. Let me make sure it is the same as in the hold notification trigger output. |
| 15:05 |
Dyrcona |
Yeah, from and to and all that, it looks like. |
| 15:15 |
tsbere |
Unless you are missing a blank line between the headers and the start of the content, causing something to barf on not-headers, I don't know without seeing the template. Or an output sample, I guess. |
| 15:16 |
mmorgan |
Dyrcona: I seem to remember having a similar problem several months ago. Can't exactly remember the resolution, but possibly the sms call numbers were using a different SMTP server than the hold notifications? |
| 15:17 |
Dyrcona |
mmorgan: I'm looking into that, but so far I'm not finding where the configuration would be different. |
| 15:24 |
Dyrcona |
The template output looks good. Similar to a hold notification: 3 header lines, a blank line, and the body. |
| 15:29 |
csharp |
Dyrcona: configuration would be the same for SMS as for email (opensrf.xml/<email_notify>/<stmp_server>) - assuming you've gotten that far already :-) |
| 15:29 |
|
rlefaive joined #evergreen |
| 15:32 |
Dyrcona |
csharp: That is what I thought, but I was scouring a few other places just to make sure. |
| 15:32 |
mmorgan |
Dyrcona: possibly we had to configure smtp server on the apaches... |
| 15:33 |
mmorgan |
I seem to remember that the call numbers were sent from the apaches while the hold notifications came from utilty. |
| 15:34 |
Dyrcona |
mmorgan: We're now using the same configuration file everywhere. |
| 15:34 |
Dyrcona |
I don't see any log entries in the mail software on the servers with apache. |
| 15:34 |
Dyrcona |
Well, other than about 1 per day for a daily cron email. |
| 15:35 |
Dyrcona |
Is it sending directly via Net::SMTP? |
| 15:35 |
* Dyrcona |
sees if he can find it. |
| 15:36 |
csharp |
Email::Send |
| 15:37 |
csharp |
/usr/local/share/perl/5.18.2/OpenILS/Application/Trigger/Reactor/SendEmail.pm (on my 14.04 box) |
| 15:37 |
csharp |
SendSMS.pm is basically a wrapper for SendEmail.pm |
| 15:41 |
Dyrcona |
csharp: Yeah. I was about to comment on that... |
| 15:43 |
Dyrcona |
So, it should run on the server that runs action_trigger runner, right? |
| 15:47 |
csharp |
we have postfix running on a separate server |
| 15:47 |
Dyrcona |
I have the default exim from Debian configured to use GMail as a smart host. |
| 15:48 |
Dyrcona |
Email notifications and SMS for holds are working.... |
| 15:49 |
jeff |
You are likely going to run into issues if you are trying to send any volume of mail through Google SMTP servers, even if you're authenticating with a G Suite account. |
| 15:49 |
jeff |
That doesn't sound like your current problem, but just a heads up. |
| 15:49 |
jeff |
s/any volume/any significant volume/ |
| 15:50 |
Dyrcona |
jeff: Right. These don't seem to get to smtp on any of the possible hosts. |
| 15:50 |
* jeff |
nods |
| 15:51 |
jeff |
for future reference, G Suite sending limits are documented here https://support.google.com/a/answer/166852?hl=en and here https://support.google.com/a/answer/2956491#sendinglimitsforrelay |
| 15:58 |
jeff |
Dyrcona: we don't use that feature, so I could be wrong, but it looks like because the A/T events in question are on-demand, they're going to be fired on whatever host handles the open-ils.trigger.event_group.fire call. |
| 15:59 |
jeff |
(or open-ils.trigger.event.fire) |
| 15:59 |
jeff |
likely the latter in this case. |
| 16:02 |
tsbere |
Dyrcona: I concur with jeff. You need to ensure your drone servers can send email. Which may not be your apache servers. |
| 16:03 |
Dyrcona |
Right. I'm just figuring that out. |
| 16:04 |
jeff |
Consult your /opensrf/default/email_notify/smtp_server config setting on the OpenSRF host(s) in question, go from there. |
| 16:04 |
jeff |
If it's localhost, MTA may be eating / queueing the mail. If it's non-localhost, you might be getting firewalled or rejected (relaying denied, whatever) -- you'd likely be seeing those rejects in your logs, so firewall/routing is more likely? |
| 16:04 |
Dyrcona |
jeff: That's the same. When we switched to GMail, I don't think I configured the drones to use a smarthost. |
| 16:05 |
Dyrcona |
I found that the test I tried to send myself was handled on drone 2 of brick 2. |
| 16:06 |
Dyrcona |
I'll check mail logs there. |
| 16:08 |
jeff |
Also, info/warn opensrf logs containing "SendEmail Reactor" |
| 16:11 |
Dyrcona |
jeff: That latter gives me what I expect: a number of entries from the drones saying that they can't send email. |
| 16:12 |
Dyrcona |
So, I need to configure the drones to use a smarthost. |
| 16:14 |
Dyrcona |
And, there's the "bounce" for my SMS frozen in the queue.... |
| 16:15 |
Dyrcona |
I need to do more than just configure a smarthost. I'll need to copy the config bits that throw out the bounces to no-reply@. |
| 16:20 |
Dyrcona |
Which also means that I'll need to find it and copy it to the right place on the drones. |
| 16:20 |
tsbere |
This is why I configured all of MVLC's production VMs to smarthost to a single machine that then sends to the outside. Only one place to handle most of that. |
| 16:24 |
|
DPearl left #evergreen |
| 16:25 |
Dyrcona |
Well, I still need to configure the smarthost on the drones. :) |
| 16:36 |
Dyrcona |
So, slightly different plan: I'll configure the server that sends the action trigger emails to relay for the local IPs, and then configure the drones to use that server as a smarthost. |
| 16:36 |
Dyrcona |
That cuts down on extra configuration to keep track of. This is, after all, being done in "the Debian way." |
| 17:00 |
pinesol_green |
News from qatests: Test Failure <http://testing.evergreen-ils.org/~live> |
| 17:04 |
|
mmorgan left #evergreen |
| 17:13 |
bmills |
anyone had luck with getting autosuggest to stop crashing on search terms beginning with punctuation? or could point me in the right direction? |
| 17:14 |
bmills |
thinking along the lines of lp 1161095 |
| 17:14 |
pinesol_green |
Launchpad bug 1161095 in Evergreen "AutoSuggest crashes when search terms begin with some punctuation" [Medium,Triaged] https://launchpad.net/bugs/1161095 |
