@ListenBrainz/@MusicBrainz android app devs, I have the MusicBrainz android app installed and I get a toast message that a "user token has not been set!" everytime I play some videos. In the app I set my LB token but I still get this message... Any idea what is happening ?
akshaaatt[m]
<Lotheric_ " @ListenBrainz/@MusicBrainz andr"> Hi Lotheric_ ! I will go through the code today regarding this and get back to you as soon as I find any bugs and try to fix it. Else we always have _lucifer to help us out.
MRiddickW has quit
sumedh joined the channel
_lucifer
ruaok: i'd take a look at it but stats.me seems to be down currentyl ;)
Lotheric_: thanks for the report. can you please also share your android version?
akshaaatt[m]: thanks for taking a look at it. let me know if you need any help.
reosarevok wonders if editor merging would be a thing we could consider (so, moving someone's edits with a second account to their main one). Someone asked support about it - I expect it's less likely to be a problem now that we block email reuse tho
sumedh joined the channel
ruaok
mooin!
_lucifer: yes, it seems that stats was down (or at least unreachable for a while) I
I'll have to harden things more still.
and the rise is because the MB mapper doesn't load old stats from disk when it starts -- I'll need to do that next.
_lucifer
ah ok, makes sense.
also, we should apply for Youtube OAuth verification today methinks
let's go over it once again others are available and then fill the verification form?
ruaok
ok, lets do that this afternoon.
sure
_lucifer: the dumps failed AGAIN, because one prerequisite temp dir was missing. grr. I've made a ticket for that and the full dump is finishing mogrifying.
and somehow, we still didn't get a dumps are outdated email. sigh.
_lucifer
damn
ruaok
in fact, cron seems to have done nothing at all overnight.
I'll let the dump finish before I investigate.
_lucifer
docker logs only shows entries from run stats crontab, but nothing from dumps crontab. ideas why that is so?
_lucifer: I saw your comment showing ruaok how to source services.sh in prod. Do you want to have a brainstorm about how we can make a 'deploy.sh' to do all of these steps
_lucifer
alastairp, sure
alastairp
take a project/service/date; pull image; stop old one; start new one; optionally update config file
ruaok
Yes, I do
alastairp
there is a script somewhere in docker-server-configs which does 90% of this
I think yvanzo wrote it, and Mr_Monkey uses it sometimes
for me, the main disadvantage of this script is that it still starts services by running start_services.sh, this means that 1) if the image isn't pulled yet, there is an extended downtime between when the service is shut down and when it starts up again after the pull, and 2) if the service we want to start is at the end of the node file it also takes a while to get to it
we were talking with ruaok and Mr_Monkey in the office a few weeks ago - one option is to try and make the `start_listenbrainz_web` function similar enough to the container name (listenbrainz-web-$DEPLOY_ENV) that we can automatically convert from one to the other
if that means that we have to rename some startup functions or some container names then that's fine
_lucifer
interesting, yeah makes sense.
alastairp
I don't know if you can call a function in bash given a string containing the name of that function. atj ?
_lucifer
i think we can.
because $VERSION errors with no command with value inside the variable found.
we could possibly even take a recommendation from your book and use python for this, just fork out to a shell for the startup. let me see how well it comes together
_lucifer
i was going to suggest that because bash is magic :).
yvanzo
eval "$function_name" # given the variable function_name contains the name of a function
_lucifer: I think we could do most of the deploy items in this document as a test, and then if it works we could get something like fabric/ansible for the final step
_lucifer
alastairp, yup makes sense.
Etua joined the channel
TOPIC: MetaBrainz Community and Development channel | MusicBrainz non-development: #musicbrainz | Channel is logged; see https://musicbrainz.org/doc/IRC for details | Agenda (May 17th is cancelled; next meeting: 2021-05-24): Reviews
alastairp
good luck to all MB deployers!
sumedh has quit
BrainzGit
[musicbrainz-server] reosarevok opened pull request #2105 (master…MBS-11664): MBS-11664 / MBS-11666: Enable discID tab for releases that wrongly have discids + add report https://github.com/metabrainz/musicbrainz-serve...
Freso
Cancel culture has gone too far. Cancelling entire days now. 😭
i think these scripts already cover a lot of the functionality we are looking for. thanks! :D
alastairp
yvanzo: of course you would have had to rewrite it at some point from fabric 2, so I guess the options were fabric 3 or bash, and you chose bash?
_lucifer
alastairp: unrelated to this but did you see the docker pull issue yestrerday? we were unable to pull metric-writer images on williams because the rate limit was exceeded. and were wondering if its finally time for our own image registry?
alastairp
_lucifer: ah, I did see that in the history.
ruaok
well, it was fixed by logging in.
alastairp
someone asked if it was logged in
I recall seeing `docker login` somewhere in some script
_lucifer
right, the issue was solved with login.
alastairp
oh, interesting. _jenkins_ does a login
ruaok
I think our own registry would be nice, but unless its less than 1 hour of work, perhaps lets do it later when we have fewer open things.
alastairp
I guess because it does lots of docker pulling
I think it'd be just as easy to ensure that we login on all nodes
in fact, much less work than our own registry
_lucifer
right makes sense.
ruaok
good short term solution. but I suspect that docker will tighten these rules over time. eventually we will want to do this.
alastairp
yeah, possible
_lucifer
there's some weird issue with the latest lb release.
ruaok
ohh, the FAIL ALL button becomes available today. eggciting!