| Time |
Nick |
Message |
| 06:54 |
JBoyer |
jeffdavis, there was an upgrade script that requested a fresh DYM build and included changed SQL to pull the values. One major difference in was that it excluded metabib values from deleted bibs. jipringle's recent bug about it suggesting numbers sounds like a rebuild with the most recent selection sql and most recent sideload script may help. |
| 07:25 |
|
kworstell-isl joined #evergreen |
| 07:42 |
|
collum joined #evergreen |
| 07:49 |
|
BDorsey joined #evergreen |
| 08:10 |
|
redavis joined #evergreen |
| 08:44 |
|
kworstell_isl joined #evergreen |
| 08:44 |
|
kworstell_isl_ joined #evergreen |
| 08:44 |
|
mmorgan joined #evergreen |
| 08:51 |
|
sleary joined #evergreen |
| 09:00 |
|
Dyrcona joined #evergreen |
| 09:27 |
Bmagic |
I'm attempting to setup an Evergreen server without any data. Just the schemas. autogen.sh is giving me this: Undefined address for Socket::pack_sockaddr_in at /usr/lib/x86_64-linux-gnu/perl/5.30/Socket.pm line 851 |
| 09:28 |
Bmagic |
I'm wondering if it needs* at least one system and one branch? |
| 09:28 |
Dyrcona |
I don't think you'd run autogen.sh with nothing in the database. |
| 09:28 |
Bmagic |
ok, fair enough, so I tried skipping that and just moving on |
| 09:29 |
Dyrcona |
You can't use Evergreen until you put some stuff in there anyway. |
| 09:30 |
Bmagic |
I'm getting internal server error, and the logs have a similar complaint: |
| 09:30 |
Bmagic |
apache2[368250]: [perl:error] [pid 368250] [client <my computer's IP>:0] egweb: Context Loader error: Undefined address for Socket::pack_sockaddr_in at /usr/lib/x86_64-linux-gnu/perl/5.30/Socket.pm line 851. |
| 09:31 |
Bmagic |
I'll try loading the DB with concerto just to make sure it's not something else |
| 09:31 |
Dyrcona |
Sure. As you said you need at least 1 org unit before anything will work. |
| 09:31 |
Bmagic |
The seed data does make CONS |
| 09:31 |
Dyrcona |
It could be something else. |
| 09:31 |
Dyrcona |
Your previous comment implied you skipped that somehow. |
| 09:32 |
Bmagic |
I skipped autogen, and just continued to stand up apache and nginx, etc. |
| 09:33 |
Dyrcona |
I guess I misinterpreted "I'm wondering if it needs* at least one system and one branch?" Yeah, "it" probably does. |
| 09:33 |
Bmagic |
CONS is there |
| 09:34 |
Dyrcona |
Errors from Socket.pm are a bit surprising if that's the problem, but it could be that some service doesn't run. |
| 09:34 |
Dyrcona |
Last time I tried this, it was for the MVLC migration in 2012... |
| 09:38 |
Bmagic |
just about to test concerto. If I get the same error, then I'll chase a different problem |
| 09:44 |
Bmagic |
same prob |
| 09:44 |
Bmagic |
I'll figure it out. This should* work. I setup concerto machines all the time |
| 09:46 |
Dyrcona |
I'd osrf_control --diagnostic and check that ejabberd, memcached, etc., are running. |
| 09:47 |
Bmagic |
too late, just deleted the machine |
| 09:47 |
Dyrcona |
Heh. That's one way to "fix" it. :) |
| 09:47 |
Bmagic |
yep, lol |
| 09:47 |
Bmagic |
that machine put me through a slog of stuff to even get that far. Time to start over |
| 09:48 |
Dyrcona |
So, what are you trying to do by creating an Evergreen installation without data? You want to start one from scratch? |
| 09:48 |
Bmagic |
yep |
| 09:49 |
Dyrcona |
I should do that just to see what happens. I set up MVLC that way, but I had SQL and other scripts to fill in their data after the initial database creation. |
| 09:51 |
berick |
Bmagic: interesting, what's the actual goal, though? |
| 09:51 |
Bmagic |
migrating a new library from another ILS to Evergreen |
| 09:53 |
* Dyrcona |
nods. |
| 09:53 |
Dyrcona |
Hm... It looks like I deleted my hz2eg repository on github.... |
| 09:55 |
berick |
i see. yeah, i'd be pretty surprised if that turned out to be easy to do |
| 09:56 |
berick |
probably lots of assumptions in the code about the presence of more than just an org unit |
| 09:56 |
Bmagic |
berick: I'm excited for the challenge. Plus, I'd like to say that I'm going to go with PG15 |
| 09:56 |
berick |
Bmagic++ |
| 09:56 |
Bmagic |
at least at first :) |
| 09:56 |
Dyrcona |
Bmagic++ |
| 09:57 |
Dyrcona |
I've been using Pg 15 with test VMs for awhile. I've even got one with produciton data. It's faster at some things than Pg 10, even when Pg 15 isn't optimized. |
| 09:57 |
Dyrcona |
We should start looking at Pg 16. It was released already. |
| 09:57 |
Bmagic |
I figured this would be an opportunity to see if Evergreen (with nothing and no history) can setup and work on PG15. I'll be watching the PG logs like a hawk |
| 09:58 |
Dyrcona |
Bmagic: It should. |
| 09:58 |
Bmagic |
I'm sure it will look* like it's working |
| 09:58 |
Bmagic |
I'll be expecting to see some nasy stuff in the logs, but hey! maybe not |
| 09:58 |
Dyrcona |
I told you it will* work. What more do you need? |
| 09:59 |
Bmagic |
running through the migration of 500k bibs into an empty DB with PG15 should be a good test |
| 10:00 |
* Dyrcona |
loads dumps and has imported bibs into Pg 15 with production data. It should work. |
| 10:01 |
Bmagic |
wouldn't it be funny if PG15 was the issue with this socket error |
| 10:01 |
Dyrcona |
Did you set pg_hba.conf properly? |
| 10:01 |
Bmagic |
I think so, I confirmed that the Evergreen application connects |
| 10:02 |
Dyrcona |
And the listen_address in postgressl.conf? |
| 10:02 |
Bmagic |
PG is configured to * for listening address |
| 10:02 |
Dyrcona |
postgresql.conf... |
| 10:02 |
Bmagic |
and 0.0.0.0/0 md5 all all pg_hba |
| 10:02 |
Bmagic |
PG15 didn't throw me off, it looks like 10 |
| 10:02 |
Dyrcona |
Nah... That's not how you do it. |
| 10:03 |
Bmagic |
I'll secure it later, I don't want that* to be my problem |
| 10:03 |
Dyrcona |
Don't use md5 with Pg 15. The default of scram whatever works. |
| 10:03 |
Bmagic |
I see, I'll change that |
| 10:04 |
Bmagic |
host all evergreen 0.0.0.0/0 scram-sha-256 |
| 10:04 |
Dyrcona |
Localhost should be allowed in pg_hba.conf by default. |
| 10:04 |
Bmagic |
my EG app isn't local |
| 10:04 |
Dyrcona |
You're allowing the entire Internet with that mask, if it even works. |
| 10:04 |
Bmagic |
right |
| 10:04 |
berick |
Bmagic: just curious, why do you need EG running to batch-load data from another ILS? will people also be manually creating data in the UI? |
| 10:05 |
Bmagic |
yeah, my co-workers are implementing the org tree |
| 10:05 |
Dyrcona |
I don't think you'll be able to implement the org tree via the client. |
| 10:05 |
Bmagic |
and circ rules, shelving locations, etc |
| 10:06 |
Bmagic |
I'll let you know |
| 10:06 |
berick |
isn't there like 1 default shelf location, circ rule? |
| 10:07 |
Dyrcona |
I found an out-of-date version of my Horizon to Evergreen migration stuff. |
| 10:07 |
Bmagic |
the admin account is stubbed into actor.usr, CONS is stubbed into actor.org_unit. stacks is stubbed into asset.copy_location |
| 10:07 |
Bmagic |
so, it should be enough to login and register a workstation and get started |
| 10:08 |
Dyrcona |
Well, you'll find out and let us know. :) |
| 10:08 |
Dyrcona |
I inserted the org. tree and a bunch of other stuff with SQL. |
| 10:08 |
Bmagic |
If I can get over these other* issues, I'll let you know how it goes |
| 10:09 |
berick |
may the schwartz be with you! |
| 10:09 |
Bmagic |
berick++ |
| 10:09 |
Dyrcona |
Heh |
| 10:09 |
Bmagic |
I just reviewed that scene on Youtube last week. LOL. Gotta love Mel Brooks |
| 10:11 |
Dyrcona |
Yeah. |
| 10:14 |
berick |
Bmagic: what about permission types, org setting types, standing penalty types, ... there's /so/ much needed data in the seed data. (I clearly can't let it go :) |
| 10:14 |
Bmagic |
I think the perm tree is there. I'm kinda familiar with this because I wrote that program to generate the enhanced concerto set |
| 10:14 |
Dyrcona |
That's some of the "other stuff" I mentioned before. :) |
| 10:16 |
berick |
e.g. the seed data adds 626 rows into config.marc21_physical_characteristic_value_map. how else do you plan to get that data? |
| 10:16 |
Dyrcona |
berick: I think he's running the seed data. He's skipping concerto. |
| 10:16 |
Bmagic |
I figured I would encounter what I encounter and solve each thing |
| 10:16 |
berick |
Bmagic: skipping seed data or concerto data? |
| 10:17 |
Dyrcona |
Bmagic: Are you using eg_db_config? |
| 10:17 |
Bmagic |
concerto. Seed is there |
| 10:17 |
berick |
ohhh |
| 10:17 |
berick |
you def. don't need concerto data |
| 10:17 |
berick |
ok, i'm way less confused now |
| 10:17 |
Bmagic |
sorry that wasn't clear, yeah, seed is there |
| 10:17 |
Dyrcona |
The confusion is probably my fault. :) |
| 10:18 |
Bmagic |
I meant without concerto |
| 10:18 |
berick |
pretty only devs, testers, etc. use concerto |
| 10:18 |
berick |
be odd to use it for a production install |
| 10:19 |
Dyrcona |
Way too much to delete.... |
| 10:19 |
Dyrcona |
:) |
| 10:21 |
Dyrcona |
Bummer. Wonder why I thought it was good idea to delete that repository? Guess I was "cleaning up" some junk. |
| 10:21 |
Bmagic |
back on pg15, It's still true that no one is using > 10 in production ? |
| 10:21 |
berick |
still on 10 here, pending some other infrastructure work |
| 10:22 |
Dyrcona |
Bmagic: I'm not sure. sharpsie was talking about jumping versions last year/early this year. |
| 10:22 |
Bmagic |
I thought I saw jeff or jeffdavis mention pg14 on their production environment but I might have been hallucinating |
| 10:22 |
Dyrcona |
I'd like to go to Pg 15 honestly, but I don't think my users are ready for the jump, yet. :) |
| 10:23 |
Dyrcona |
We'd also have to back port some DB changes to production before we could go to anything higher than Pg 11. |
| 10:23 |
Bmagic |
but EG 3.11.1 has what it needs? |
| 10:24 |
Dyrcona |
yeah, I think it does. |
| 10:24 |
Dyrcona |
If it has the installation for Pg 15, then it should have all of the commits. |
| 10:25 |
* Dyrcona |
makes a Lp bug.... |
| 10:27 |
Bmagic |
I'm testing my theory that my socket issue was due to the Linux host being 22.04 and the container was 20.04 |
| 10:28 |
Dyrcona |
Maybe..... |
| 10:28 |
Dyrcona |
You can run Evergreen on 22.04. It works there, too. |
| 10:28 |
Bmagic |
yeah, I might target 22.04 for this. I can't remember why I shy away from that one. There was something about it but I can't remember |
| 10:28 |
Dyrcona |
I do most of my testing on Ubuntu 22.04. |
| 10:29 |
Dyrcona |
Lp 2037656 |
| 10:29 |
pinesol |
Launchpad bug 2037656 in Evergreen "Add Support for PostgreSQL 16" [Wishlist,New] https://launchpad.net/bugs/2037656 |
| 10:29 |
Bmagic |
Dyrcona++ |
| 10:30 |
Dyrcona |
I think a number of us in the community have an aversion to using the latest and greatest in production, preferring to stick with the tried and true. |
| 10:31 |
Bmagic |
I've gotten that way too. I remember when I first started, I was installing EG on Fedora |
| 10:32 |
Bmagic |
I learned the error in my ways and moved over to Ubuntu and things just worked better, so, yeknow |
| 10:32 |
Dyrcona |
Well, Fedora is/was a fast-moving distro. You can get in trouble if you try non-LTS Ubuntu releases, too. |
| 10:52 |
Stompro |
Going live with PG13 on Oct 7th. |
| 10:55 |
Dyrcona |
Stompro: Any particular reason to choose 13 and not 14 or 15? |
| 11:04 |
|
briank joined #evergreen |
| 11:07 |
Stompro |
I ran into an upgrade issue where an import from our current system running on 9.6 failed in 14 and 15. I'll plan on moving up to 14 or 15 in a few months. |
| 11:08 |
Stompro |
We are going to 3.11.1 on the 7th also. |
| 11:09 |
Dyrcona |
I was about to say that is probably patched in 3.11, but I think you'd have to apply the particular update to the old db before dumping and restoring. |
| 11:22 |
|
kworstell_isl joined #evergreen |
| 11:36 |
jmurray-isl |
I'm currently troubleshooting an issue that's occurring on only 1 of my instances (out of 5) where folders using OpenILS::WWW::Proxy::Authen are claiming too many redirects. I've compared the apache and nginx configurations, along with the installed perl modules, and they are all the same. If I change the eg_vhost folders that use these to a configuration that uses OpenILS::WWW::Proxy, instead, it |
| 11:36 |
jmurray-isl |
works. Does anyone have any immediate thoughts on what might cause this? |
| 11:55 |
Bmagic |
jmurray-isl: my knee jerk reaction is: delete the vm and make a new one. Assuming you have a way to do that easily and reliably |
| 11:56 |
jmurray-isl |
Bmagic: I'm tempted, but this is a bare metal box, using a full copy of our production database. ;) |
| 11:58 |
Bmagic |
the Evergreen application is installed on the same machine with PG? |
| 11:58 |
Bmagic |
when you said 1 machine (out of 5) I assumed each one had a single-purpose dedicated OS |
| 12:01 |
Bmagic |
how about this: what if you setup a new brick from scratch. Does the problem occur there? |
| 12:01 |
jmurray-isl |
Yes, we have two non-prod bare metal machines that use a full copy of the prod database, one for migration testing, and the other for practice. I eventually plan to move practicing over to the migration testing server, but currently it helps to have them separate. The others would be training/testing VMs, which work fine. |
| 12:03 |
jmurray-isl |
At this point, it would probably save time just to rebuild the server in question, but I wasn't sure if there'd be a quicker solution. |
| 12:04 |
Bmagic |
if you can't find a difference between the working machine and the non-working machine. A difference in any of the meaningful Evergreen configs and Apache configs, then it's gotta be OS |
| 12:04 |
jmurray-isl |
One would think. The OS should be the same (both Ubuntu Focal). |
| 12:05 |
Dyrcona |
jmurray-isl: Does the too many redirects message prevent it from working, or does it go away and then things work? |
| 12:05 |
Bmagic |
(or rather, more specifically, other things that you're not diffing, whcih include auxillary perl mods maybe which might be due to a slight difference in version of the OS) |
| 12:06 |
jmurray-isl |
Dyrcona: It prevents it from working. |
| 12:07 |
Dyrcona |
Reloads in the browser don't help? I've seen "too many redirects" from time to time, but it eventually works. |
| 12:08 |
jmurray-isl |
Reloads don't help, and neither does Empty Cache and Hard Reload. |
| 12:09 |
|
jihpringle joined #evergreen |
| 12:15 |
Bmagic |
Dyrcona berick: check this out: https://pastebin.com/8FqRimX4 |
| 12:16 |
Bmagic |
I suppose that doesn't stop the 950 seed from continuing? |
| 12:17 |
Bmagic |
it's this line |
| 12:17 |
Bmagic |
INSERT INTO biblio.record_entry VALUES (-1,1,1,1,-1,NOW(),NOW(),FALSE,FALSE,'','AUTOGEN','-1','<record xmlns="http://www.loc.gov/MARC21/slim"/>','FOO'); |
| 12:17 |
Dyrcona |
"Ingest of biblio.record_entry -1 failed" Yes. That record is empty, there's nothing to ingest. |
| 12:17 |
jeffdavis |
Bmagic: haven't read scrollback yet but yea we're on PG14 (with EG3.9, upgrading to 3.11 in a few weeks) |
| 12:18 |
Bmagic |
jeffdavis: I wasn't hallucinating! |
| 12:18 |
Bmagic |
jeffdavis++ |
| 12:19 |
Bmagic |
Dyrcona: Yeah, I was worried that it killed the seed execution, but I guess not. Moving on |
| 12:19 |
Dyrcona |
Bmagic: I've seen some messages pop up with eg_db_config on recent versions of PostgreSQL lately. I've made a mental note to look into them, but haven't really done much. I recall mentioning it in channel once and someone else has also noticed. |
| 12:19 |
Dyrcona |
The upshot is there are things to look into, but nothing fatal so far. |
| 12:20 |
Dyrcona |
jmurray-isl: If OpenILS::WWW:Proxy is working for you, then switch to that module. I'm not sure that the Authen module adds much of use. |
| 12:29 |
Dyrcona |
jmurray-isl: Just curious, are these local, custom directories or is the stock Evergreen configuration using OpenILS::WWW::Proxy::Authen? |
| 12:31 |
Dyrcona |
Oh. I didn't realize that many places use Proxy::Authen. |
| 12:43 |
|
bgillap joined #evergreen |
| 13:12 |
jmurray-isl |
Dyrcona: They are stock eg_vhost, such as "/reporter" and "/standalone" |
| 14:03 |
JBoyer |
jmurray-isl, this sounds vaguely familiar, could nginx be redirecting things in addition? |
| 14:35 |
berick |
Bmagic: playing The Resistance :) https://xkcd.com/2833/ |
| 14:36 |
Bmagic |
berick++ lol |
| 14:36 |
Dyrcona |
Heh. |
| 15:00 |
|
mmorgan1 joined #evergreen |
| 15:27 |
redavis |
We all know. Bmagic is a spy. |
| 15:30 |
|
collum joined #evergreen |
| 15:42 |
|
jihpringle joined #evergreen |
| 15:49 |
Bmagic |
redavis: or am I? |
| 15:49 |
redavis |
always. the question is...a spy or a double agent? |
| 15:50 |
Bmagic |
Think about this: a triple* spy |
| 15:50 |
redavis |
also ty to berick: I'm closing out this very full day looking at xkcd comics. |
| 15:50 |
Bmagic |
It's exactly the same as "6 minute abs" - we offer the same work out as the 7 minute folks, but we do it 1 minute faster |
| 15:50 |
redavis |
lol, a triple spy is just a plain ol' spy that likes chaos. |
| 15:53 |
Dyrcona |
redavis: Kim Philby. |
| 15:55 |
redavis |
oy, I mean, chaos. |
| 15:55 |
Dyrcona |
The Chao that can be understood is not the true Chao. |
| 15:56 |
redavis |
I wonder if he ever thought "you know what'd be fun? gardening and walking around aimlessly in a safe wooded environment." |
| 15:56 |
redavis |
or "today, I'll look at birds and converse with chickadees and not add fuel to war fires." |
| 15:56 |
Dyrcona |
That generation had such interesting lives. Roald Dahl was another one. |
| 15:57 |
redavis |
thank goodness we're such a tidal creature. |
| 15:58 |
redavis |
And, on that note, I'm gonna go look at birds and converse with chickadees. Maybe walk around aimlessly somewhere vaguely resembling outside. |
| 15:58 |
redavis |
y'all be safe. |
| 16:03 |
jeffdavis |
Declare by Tim Powers has a pretty entertaining explanation for Kim Philby-related chaos. |
| 16:05 |
Dyrcona |
jeffdavis++ That looks interesting. I may have to read ti. |
| 16:45 |
jmurray-isl |
JBoyer: The nginx config is taken directly from the other bare metal machine (adjusting for domain), which works perfectly fine. |
| 17:14 |
|
mmorgan left #evergreen |
| 20:01 |
|
sleary joined #evergreen |
| 20:04 |
|
jihpringle joined #evergreen |
