#metabrainz

the main drive is resyncing (probably because of reallocated bad sectors)

this is going to affect the speed of the dumps (and the dumps will affect the speed of the resync)

12 days to finish the resync

okay, so they should be a bit faster on frank then.

what disks does frank have?

let me check

I'm going out for a bit, but LordSputnik, pong me when you have a test ready for me :D

this is the server: https://www.hetzner.de/dedicated-rootserver/ex51

right, normal spinning disks, then

yes

7200 rpm

:/

4TB is SSD is quite expensive. :(

*of

I can imagine

lolyea

another 30GB file done and being compressed rn

mmm. faster??

this is the second one, I don't think so, I think it took around the same time.

Mr_Monkey: LordSputnik https://gist.github.com/shivam-tripathi/1e60558...

This gist is from https://api.travis-ci.org/v3/job/403097211/log.txt

I think postgres version is causing issue here.

Later tests which fail are:

`insert into "bookbrainz"."editor_type" ("id", "label") values ($1, $2) returning "id" - relation "bookbrainz.editor_type" does not exist[0m[90m error: relation "bookbrainz.editor_type" does not exist`

I wonder if transaction is aborted then how partial tables are created and remaining tests pass

Postgres version is 9.6.6, so something else is perhaps the matter.

CatQuest: won't be today I'm afraid, had guests over last night and I'm out tonight - but the code is working now and I just need to commit everything and update the server, so maybe I can do that tomorrow night

:(

bukwurm Thanks, I'll have a look!

bukwurm & LordSputnik : It looks like a CI-breaking change was introduced to bookbrainz-sql with this commit, and it looks like no job passed since https://github.com/bookbrainz/bookbrainz-sql/co...

I can't figure out where the postgres configuration is in Travis. I think maybe the version 9.6.6 that bukwurm saw is only that of psql rather than the postgres server version, in which case it might just need an update to >9.4

Mr_Monkey: Yes, the gist shared showed the cause of error being jsonb type missing

*I shared

jsonb was used to store metadata related to imports

jsonb support I think is in 9.6+ versions

*9.4+

bukwurm Yes, just confirming and doing a recap. Do you have access to the Travis config?

Mr_Monkey: No :/

Mr_Monkey: But I do wonder, if the entire schema creation is done in a single transaction, when it fails how come other tests pass?

Maybe something to do with tests rather than database setup I guess

I'm not familiar enough with the setup yet to venture an educated guess. It looks to me like the schema creation only fails for the last few operations though

Yeah, I saw that configuration, but there's nothing there that describes the postgres server, so I assume there is some extra config somewhere.

Mr_Monkey: I am guessing "npm run-script test-ci" is the one

That's in the bookbrainz-site package.json, and here is what it runs:

nyc --reporter=lcovonly --reporter=text-summary npm run test

That's not it then :P

alastairp: 4 files done now.

bukwurm: I'm used to specifying a postgres version in the travis file: https://github.com/MTG/freesound/blob/master/.t..., not sure why bookbrainz doesn't have it

Could it be just that? Thanks alastairp, I'll try that.

(not a travis expert)

alastairp: Ok. Could there any other way to specify postgres version on Travis?

no idea, sorry

Or for that matter, any other service.

alastairp: Ok

Mr_Monkey: As you're trying it, do tell what happens when you mention the version in the .travis.yml file

:)

Will do

👍

Mr_Monkey: looks like 'addon' is deprecated, look at https://docs.travis-ci.com/user/database-setup/...

"The Travis CI build environments use version 9.2 by default on Trusty images, but other versions from the official PostgreSQL APT repository are also available."

That's gotta be it, I think.

bukwurm & alastairp Boom chacka boom ! all tests passing. Thanks for your help !

Mr_Monkey: Awesome

Project bookbrainz-site build #35: errored in 3 min 7 sec: https://travis-ci.org/shivam-tripathi/bookbrain...

Mr_Monkey: I think getting webpack integration with module replacement is awesome for the project, it's gonna make development so much easier.

Awesome work 👍

It's quite nice indeed ! Currently the recompilation is a bit long for my taste, but I think I can optimize it.

Using Webpack for production builds I'm still working on, there are a few snags, especially with css/less.

And thanks :)

bukwurm, Mr_Monkey: nice work guys. I did think I specified postgres 9.4 - perhaps for data-js

Yeah: bookbrainz-data-js/blob/master/.travis.yml

* https://github.com/bookbrainz/bookbrainz-data-j...

We just need to do the same for the site repo

LordSputnik: I think the confusion due to earlier tests failures due to two versions running complicated the issue

It only today when I really looked at the logs, to find the jsonb issue

LordSputnik: Do you recall what was the issue earlier, when you mentioned something about two version causing race condition or something?

bukwurm: oh, that turned out to be a missing transaction passed to a bookshelf query

So it was running outside the transaction

LordSputnik: Ok

ruaok: see Doug's email, might be interesting :)

Sounds like Doug should be put in contact with the people who previously talked about a similar project, maybe?

Chris Mendez and Mickaël Arcos.

The latter wrote a blog post on the subject: https://deezer.io/matching-albums-through-cover...

I can send him that at least :)

I can also create a follow-up e-mail introducing the two others maybe?

Feel free! :)