Improvements to dbjobqueue

[lzap]

Couple of commits:

• Because slogger was imported from internal package, it was not possible to import jobqueue effectively. Go was erroring out, the change moves the simple logger package to pkg/ and also moves the interface into a separate file where it belongs. Both noop and logrus implementations were moved into own packages so it's now nice and clean - these can be imported without extra dependencies and the interface itself can be also imported without logrus dependency.
• Another small commit allows passing in an existing pgxpool so we can reuse the one we already have in our app. We migrated from database/sql to the native pgx last week.
• Finally, when I was testing this, I needed to create a migration for the tests so I am dropping a Makefile target to do that so others don't need to figure it out

Edit: Initially the PR had some changes around context passing, but this is worth looking into a separate PR I think.

Cheers!

[lzap]

I am gonna drop the context change, that's really weird, all functions in the dbjobqueue interface should really have a ctx parameter.

[lzap]

What's really weird is the dbjobqueue tests passed the first time locally, now I am getting error inserting the job's heartbeat: ERROR: insert or update on table "heartbeats" violates foreign key constraint "heartbeats_id_fkey" (SQLSTATE 23503) even if I drop the db and recreate it from scratch.

Looks like this is passing on the CI. Weird stuff.

[lzap]

Rebased, I've also implemented the suggested CI change.

[lzap]

Looks like some checks are stuck, shall I rebase once again?

[croissanne]

Looks like some checks are stuck, shall I rebase once again?

Github is actually throwing up an error, there needs to be a - in front of the run. (or - name: ...\n)

[lzap]

Sure, rebased. Crossing my fingers :-)

[lzap]

And once again, Mr. Schutz had some issues, but the page was 500 so not sure.

[croissanne]

And once again, Mr. Schutz had some issues, but the page was 500 so not sure.

The CI takes quite long, and often has at least 1 or 2 infra flakes. Easier to just retry the individual jobs.

[lzap]

I dont understand, Schutzbot job reports an error, when I click on the link it says it's running.

And the RPM, oh, I almost forgot how bad rpmbuild is with reporting errors. It fails but I am not following why.

[croissanne]

I dont understand, Schutzbot job reports an error, when I click on the link it says it's running.

And the RPM, oh, I almost forgot how bad rpmbuild is with reporting errors. It fails but I am not following why.

fedora37 ppc and centos tream 9 are also failing on other PRs; it's unrelated. I'll keep retrying schutzbot and merge once green

[lzap]

Ok let me know if I need to rebase or something. Thanks.

[lzap]

It looks like our tasking system will be more simple than I thought, we are aiming to use very simple "deffered function call" style of execution and we ended up not using dependant jobs at all, therefore, I am closing this PR. Thank you for the effort, apologies for that you worked on extracting the jobqueue to pkg/ it looks like it won't be used at the moment.