| Time |
Nick |
Message |
| 07:42 |
|
BDorsey joined #evergreen |
| 07:47 |
|
Rogan joined #evergreen |
| 08:16 |
|
dguarrac joined #evergreen |
| 08:26 |
|
mantis1 joined #evergreen |
| 08:36 |
|
kmlussier joined #evergreen |
| 08:41 |
|
sleary joined #evergreen |
| 08:44 |
kmlussier |
Good morning #evergreen! |
| 08:50 |
|
Dyrcona joined #evergreen |
| 09:03 |
|
terranm joined #evergreen |
| 09:09 |
sleary |
hi kmlussier! |
| 09:14 |
|
terranm joined #evergreen |
| 09:16 |
|
mmorgan joined #evergreen |
| 09:18 |
kmlussier |
sleary: I hope you enjoyed your first hack-a-way! |
| 09:21 |
Dyrcona |
So, for those following the marc_export saga: I'm logging the time spent in subroutines in microsecond granularity using Time::HiRes. I hope this points to where the trouble is. I'm also only exporting 1 library's items. They have about 145,000 records so it should finish in a "reasonable" amount of time. |
| 09:26 |
|
redavis joined #evergreen |
| 09:27 |
Dyrcona |
berick: I'm going to do some more research on Rust and look into adding a debug option to the Rust marc-esport.rs. I'd like to log the queries it runs, because it looks like the main query ran for 5 secs each time it was called on my produciton db the other day. I want to get the whole thing to run through explain analyze. It gets truncated in the postgres logs. |
| 09:37 |
Dyrcona |
Rogan left. I wanted to ask him about some queries he wrote, i.e. why he used subqueries instead of join in some places. Is it because the subqueries were faster or was that just how it came to him? I have seen subqueries be faster than joins in some cases, but not always. |
| 09:38 |
Dyrcona |
I suppose I could test it on my data and adjust if appropriate. It often depends on indexes and number of rows. |
| 09:49 |
|
terranm67 joined #evergreen |
| 09:51 |
csharp_ |
Dyrcona: I'm going to make a separate bug for my SFTP reimplementation so we can at least have functional SFTP (which currently doesn't work without my change or something like it) |
| 09:51 |
|
Rogan joined #evergreen |
| 09:52 |
csharp_ |
I'll reference the new bug from your "make EDI smarter" bug and start looking more closely at yours once I'm back home |
| 09:55 |
|
Rogan joined #evergreen |
| 09:55 |
Rogan |
Dyrcona - missed you message but someone said you had a question? |
| 09:57 |
Dyrcona |
csharp_: Sounds good. |
| 09:58 |
Dyrcona |
Rogan: Yes, I wondered if you used subqueries in the Aspen export scripts rather than joins because subqueries ended up being faster, or if that was just how the query came into your head. I know that subqueries are sometimes faster than joins depending on circumstances. |
| 09:58 |
mmorgan |
collum++ # First commit!! |
| 09:59 |
terranm67 |
collum++ |
| 09:59 |
Dyrcona |
My plane boards in about 30 minutes, so I'll be signing out soon. |
| 09:59 |
kmlussier |
collum++ |
| 09:59 |
Dyrcona |
collum++ |
| 10:00 |
terranm67 |
Safe flight Dyrcona! |
| 10:01 |
Dyrcona |
Thanks! |
| 10:01 |
Dyrcona |
I wish safe travels to everyone else! |
| 10:02 |
eeevil |
berick: I'm looking at opensrf/redis anew via the collab branches. I have two questions that jump out at me: is the --reset-message-bus documented beyond "you must do this at redis restart" somewhere? (the commit it comes in with isn't obvious to me; also, I want to help remove any instances of boilerplate stuff that we can automate or fold into other actions as early as possible -- ideally before merge); and, while the config and password list files |
| 10:02 |
eeevil |
don't require it themselves, it /looks/ like the usernames are pinned as "opensrf", "router", and "gateway" in the code that implements each part. is that correct and do you see any reason we /can't/ replace those strings with the value pulled from opensrf_core.xml as with ejabberd? |
| 10:02 |
eeevil |
(not expecting full-on answers ASAP! just want to make sure the questions are out there) |
| 10:04 |
Dyrcona |
FWIW, the --reset-message-bus thing bothers me, too, and berick promised making it more automatic at some point, but I don't think it has happened, yet. |
| 10:04 |
* Dyrcona |
signs out to find a restroom and prepare to board. |
| 10:04 |
Dyrcona |
Catch y'all later! |
| 10:05 |
abneiman |
eeevil: berick left for home about 30 minutes ago so you're probably not going to hear from him today :) |
| 10:08 |
|
sandbergja joined #evergreen |
| 10:08 |
Rogan |
Dyrcona - in that case I used subqueries just because they were easier to read, sometimes the planner does like a subquery more than a join but I didn't see a speed issue that made me experiment there |
| 10:10 |
pinesol |
News from commits: LP2002693 Remove forced reload to avoid endless logout loop <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=f3ef278b0c9f0bdbb9d86e5ba0180bd9e4ef75a2> |
| 10:27 |
terranm67 |
Made my first mistake committing a branch, but JBoyer said I didn't actually break anything - phew |
| 10:28 |
terranm67 |
I was afraid I was going to have the shortest lived term as a core committer in history |
| 10:28 |
sandbergja |
terranm67++ # it's all part of the process! |
| 10:30 |
abneiman |
terranm67++ |
| 10:30 |
JBoyer |
terranm67++ ! |
| 10:33 |
kmlussier |
terranm67: We all make mistakes when commtting. It's expected. |
| 10:36 |
|
briank joined #evergreen |
| 10:37 |
|
mmorgan left #evergreen |
| 10:41 |
pinesol |
News from commits: Merge branch 'main' of git.evergreen-ils.org:Evergreen <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=b95780d3cedaadf5d1f5e6258980aab73f0bb8a9> |
| 10:41 |
pinesol |
News from commits: LP2040184 Patron survey answer formatting <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=0344a9350d4be4fd2b909d1d63799edea868f4d7> |
| 10:42 |
Stompro |
I wonder if the marc_export performance problems come down to using insert_grouped_field to add the 852... Seems like the performance of that would get worse for each field added since it tries to insert in a certain order. |
| 10:43 |
Stompro |
Maybe one call to insert_fields_ordered with all the 852 fields at once would work better. |
| 11:41 |
pinesol |
News from commits: LP1772614 Untranslatable Strings in Billing Print Templates <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=56fbbef72a55de178266f1e9a9be0764802389a7> |
| 11:52 |
|
jihpringle joined #evergreen |
| 12:11 |
|
mmorgan joined #evergreen |
| 13:32 |
|
mmorgan left #evergreen |
| 13:55 |
|
jihpringle joined #evergreen |
| 14:02 |
|
jihpringle joined #evergreen |
| 14:38 |
eeevil |
Stompro: I would personally be surprised if that made a noticeable difference in speed overall, but ... I have been surprised before! :) |
| 14:39 |
eeevil |
(though I guess if you're Dyrcona and have a bib with 10k items attached... well, maybe?) |
| 14:39 |
Stompro |
eeevil, you are correct, it had a very small improvement. |
| 14:54 |
|
mantis1 left #evergreen |
| 14:58 |
|
jihpringle joined #evergreen |
| 16:34 |
|
sleary joined #evergreen |
| 18:13 |
|
kmlussier left #evergreen |
