| Time |
Nick |
Message |
| 06:01 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 07:21 |
|
rjackson_isl_hom joined #evergreen |
| 07:32 |
|
Dyrcona joined #evergreen |
| 07:38 |
|
rfrasur joined #evergreen |
| 07:39 |
|
rfrasur joined #evergreen |
| 07:47 |
|
mantis2 joined #evergreen |
| 07:54 |
Dyrcona |
Been a while since I ran git pull: Your branch is behind 'origin/master' by 1764 commits, and can be fast-forwarded. |
| 08:30 |
|
alynn26 joined #evergreen |
| 08:31 |
Dyrcona |
csharp: It might interest you to know that I just got better profile numbers from a default installation of Pg10 with the same authority update on the same hardware than I got last Friday with a tuned Pg 9.6 database. I'm going to try Pg 11 and Pg 12 after restoring dumps to them. |
| 08:33 |
Dyrcona |
Granted, this is totally nonscientific testing. I'm just running the update, reverting it, and running the update again with plprofiler and looking at the output. I'm not doing enough trials nor averaging results. |
| 08:40 |
csharp |
thanks! I can do some experimentation too |
| 08:42 |
Dyrcona |
I'm restoring to Pg 11, now. My methodology is do the restore, then run the plprofiler as mentioned above. I'm trying not to use the db server for anything else in the meantime. |
| 08:42 |
Dyrcona |
The restore usually takes so long that I don't get around to running plprofiler until the next morning. |
| 08:49 |
|
mmorgan joined #evergreen |
| 08:58 |
|
sandbergja joined #evergreen |
| 08:58 |
Dyrcona |
Funny how autovacuum kicks in during a restore. |
| 08:59 |
|
collum joined #evergreen |
| 09:00 |
csharp |
we usually disable it during that kind of thing |
| 09:00 |
csharp |
starts chewing up I/O |
| 09:04 |
Dyrcona |
You turn it off in the configuration, I guess. |
| 09:04 |
Dyrcona |
I didn't think of it. |
| 09:05 |
Dyrcona |
I'll try that with the restore to Pg 12. |
| 09:12 |
Dyrcona |
Fun thing: If you run more than one PostgreSQL cluster on a server, then systemctl restart postgresql restarts them all. You have to use pg_ctl to start and stop individual clusters. |
| 09:40 |
|
sandbergja joined #evergreen |
| 09:55 |
|
jvwoolf joined #evergreen |
| 10:01 |
jeff |
Dyrcona: i tend to use systemctl restart postgresql clustername.service |
| 10:01 |
|
jvwoolf1 joined #evergreen |
| 10:01 |
jeff |
though at least on Debian there used to be some annoying bugs with cluster names containing one of _ or - (I don't remember which). |
| 10:04 |
|
jihpringle joined #evergreen |
| 10:05 |
Dyrcona |
jeff: I'm not even sure that I'm using "cluster" correctly. :) |
| 10:06 |
Dyrcona |
And, this is cool: 6281 postgres: 11/main: parallel worker for PID 4631 |
| 10:12 |
Dyrcona |
jeff: I just have the postgresql@.service script. Guess I'll learn how to make the ones for each cluster. |
| 10:13 |
|
alynn26_away joined #evergreen |
| 10:16 |
Dyrcona |
OIC....It "just works" with the right thing after the @. |
| 10:16 |
Dyrcona |
Score one for systemd. :) |
| 10:16 |
Dyrcona |
jeff++ |
| 10:17 |
|
dbwells_ joined #evergreen |
| 10:25 |
csharp |
I've mostly come around on systemd - there are things I dislike about it but that's true of sysV init too |
| 10:32 |
Dyrcona |
I recently read a long blog post by someone not affiliated with systemd that takes a look at the history of where systemd comes from, what problems it was initially trying to solve, and the technical aspects of the implementation so far. |
| 10:33 |
Dyrcona |
Essentially, I am ambivalent about systemd. |
| 10:34 |
Dyrcona |
I don't really hate it any more, and I don't like it. I tolerate it. |
| 10:50 |
|
dbwells joined #evergreen |
| 11:11 |
mmorgan |
Has anyone done an action trigger custom filter that looks at a patron address field? |
| 11:38 |
Dyrcona |
I haven't, but if you add actor.usr_address to the environment, whatever the object id is, it should work. |
| 11:39 |
csharp |
we're looking at soft boundaries for holds - can anyone comment as to whether you use it and whether it works well for you? |
| 11:39 |
csharp |
on its face, I'm not sure why we've not been using them because it looks like what our libraries expect anyway |
| 11:43 |
jeff |
I'm pretty sure there are at least some libraries here that have a practice of having multiple active email addresses on a patron account. Anyone care to raise their hand or share success/pitfalls? |
| 11:44 |
jeff |
csharp: if someone other than you had asked about soft boundaries I would have pointed them in your direction. I thought that you used them. |
| 11:44 |
jeff |
(or maybe I'm thinking stalling, which may or may not be closely related...) |
| 11:49 |
csharp |
yeah, stalling is set, but I'm pretty sure that's no longer necessary and probably causes more problems than it solves |
| 11:50 |
csharp |
it was to address a courier glut that has been resolved since |
| 11:50 |
csharp |
(though who knows how the pandemic will affect things as we reopen) |
| 12:01 |
|
rjackson_isl_hom joined #evergreen |
| 12:16 |
mmorgan |
Dyrcona: Thanks, I'll give it a try. |
| 12:21 |
mmorgan |
csharp: jeff: How is a soft boundary supposed to work? |
| 12:23 |
csharp |
mmorgan: the idea is that it forces the hold targeter to restrict targeting to the selection_ou (or pickup_lib), and only look elsewhere up the hierarchy if it doesn't come up with a suitable copy |
| 12:23 |
csharp |
as opposed to "all orgs are fair game equally" |
| 12:24 |
* csharp |
glances at miker in case he's got it wrong :-) |
| 12:27 |
mmorgan |
Ok, thanks for the concise description |
| 12:27 |
mmorgan |
csharp++ |
| 12:30 |
jeff |
csharp: that matches my limited understanding. |
| 12:50 |
|
jihpringle joined #evergreen |
| 12:55 |
|
khuckins joined #evergreen |
| 13:05 |
jeff |
I've been thinking about adding a "email me my username(s)" feature. |
| 13:06 |
jeff |
so, given an email address, we would send an email with a list of all usernames associated with that email address. |
| 13:06 |
jeff |
lots more to it, of course. |
| 13:07 |
jeff |
needs to have a way to limit by org boundaries, option to avoid sending usernames that are card numbers by auto-generating a username, etc. |
| 13:08 |
jeff |
rate limiting, decision to send a "no usernames are associated with this email address" or just *not* send an email in that case, etc. |
| 13:08 |
jeff |
would others here be interested in such a thing? |
| 13:12 |
|
jvwoolf joined #evergreen |
| 13:14 |
|
Dyrcona joined #evergreen |
| 13:22 |
jeff |
and... support for finding accounts by email with multiple email addresses in actor.usr.email... |
| 13:32 |
|
jihpringle joined #evergreen |
| 14:11 |
|
jvwoolf joined #evergreen |
| 15:35 |
|
mantis2 left #evergreen |
| 16:22 |
jeffdavis |
I guess if format.date is deprecated, we can probably stop retrieving the setting every time a patron record is retrieved. |
| 16:25 |
jeffdavis |
Hmm, actually it looks like egDateInput still uses that setting? |
| 16:53 |
jeff |
yeah, my rule of thumb for any "deprecated and should be noted as such / unused and should be removed" is along the lines of "is it really, though?" :-) |
| 17:02 |
jeff |
no takers on the "I forgot my username, please send it to me in an email" feature? |
| 17:11 |
jeffdavis |
It seems useful and reasonable, I'm just not aware of any particular demand for it here. |
| 17:11 |
jeffdavis |
I do think it would be a good thing. |
| 17:14 |
jeff |
often our fallback is "your card number always works in place of your username if you've forgotten your username", but there are cases where that isn't as helpful. |
| 17:29 |
|
mmorgan left #evergreen |
| 18:00 |
|
sandbergja joined #evergreen |
| 18:01 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
| 19:31 |
|
sandbergja joined #evergreen |
| 19:51 |
|
sandbergja joined #evergreen |
| 20:23 |
|
sandbergja joined #evergreen |
| 20:34 |
|
sandbergja joined #evergreen |
| 21:01 |
|
sandbergja joined #evergreen |
| 21:07 |
|
sandbergja joined #evergreen |
| 21:27 |
|
sandbergja joined #evergreen |
| 22:03 |
|
sandbergja joined #evergreen |
| 23:01 |
|
dbwells_ joined #evergreen |
