Hmm. I've been looking at the ideas pages. I think I'll participate under BookBrainz.The documentation project looks interesting to me :) Mr_Monkey
2021-03-10 06915, 2021
_lucifer
we currently do not have any ideas regarding spark, could probably remove that?
2021-03-10 06936, 2021
ruaok
yeah, ok.
2021-03-10 06903, 2021
yvanzo
thanks, I did not know it was limited to 5 tags
2021-03-10 06950, 2021
iliekcomputers
could i get someone to deploy LB master to beta web?
2021-03-10 06950, 2021
CatQuest has quit
2021-03-10 06932, 2021
ruaok
iliekcomputers: gimme a few, then I can.
2021-03-10 06943, 2021
iliekcomputers
thanks!
2021-03-10 06929, 2021
Cyna[m] joined the channel
2021-03-10 06941, 2021
CatQuest joined the channel
2021-03-10 06941, 2021
CatQuest has quit
2021-03-10 06941, 2021
CatQuest joined the channel
2021-03-10 06900, 2021
ruaok
ok, the usual mentors for GSoC are invited. If you think I missed you, plz ping me.
2021-03-10 06938, 2021
reosarevok
ruaok: I didn't get an invite mail or anything yet, should I have? :)
2021-03-10 06956, 2021
reosarevok
(I doubt I'll end up mentoring because we generally don't have enough MB interest, but)
2021-03-10 06912, 2021
SothoTalKer joined the channel
2021-03-10 06903, 2021
SothoTalKer_ has quit
2021-03-10 06908, 2021
ruaok invites reosarevok too
2021-03-10 06903, 2021
zas
what do we do in rust?
2021-03-10 06932, 2021
ruaok
there is a proposal for some rust bindings for MB
2021-03-10 06959, 2021
CatQuest
zas: stand aroudn ridgetly and red
2021-03-10 06904, 2021
CatQuest
:D
2021-03-10 06928, 2021
reosarevok
ruaok: any chance to invite my meb account instead, so that anything related to this comes to my work inbox?
2021-03-10 06933, 2021
reosarevok
Or does it require gmail specifically?
2021-03-10 06905, 2021
ruaok
I only invited @metabrainz accounts.
2021-03-10 06925, 2021
Mr_Monkey
Hi c1e0 ! Nice to see you back here!
2021-03-10 06959, 2021
Mr_Monkey
We've been using #bookbrainz for BB-related discussions, join us there ! :)
2021-03-10 06906, 2021
c1e0
Mr_Monkey: Haha. I've always been around. Just silent :)
2021-03-10 06908, 2021
c1e0
Okay!
2021-03-10 06919, 2021
goldenshimmer joined the channel
2021-03-10 06920, 2021
lorenzuru joined the channel
2021-03-10 06920, 2021
SamThursfield[m] joined the channel
2021-03-10 06920, 2021
kepstin joined the channel
2021-03-10 06920, 2021
joshuaboniface joined the channel
2021-03-10 06902, 2021
ruaok
iliekcomputers: beta updated
2021-03-10 06901, 2021
iliekcomputers
thanks!
2021-03-10 06952, 2021
iliekcomputers
the new icons look good Mr_Monkey ! :D
2021-03-10 06912, 2021
c1e0_ joined the channel
2021-03-10 06957, 2021
reosarevok
ruaok: oh, weird, seems the system is confused. On it
2021-03-10 06916, 2021
zas
bitmap: disk space on paco is still lacking, can you please move pg-williams to ... williams?
2021-03-10 06942, 2021
ruaok
lol, too much sense.
2021-03-10 06930, 2021
c1e0 has quit
2021-03-10 06905, 2021
reosarevok
Seems whatever email you're logged in as when you click the invite link counts as "invited", at least at first. Weird. Anyway, solved
2021-03-10 06927, 2021
atj
thanks for the pinging me on that PR _lucifer and alastairp. I left a couple of comments last night, nothing major.
2021-03-10 06900, 2021
Mr_Monkey
Kewl !
2021-03-10 06958, 2021
ruaok
zas: how are we doing on replacing the gateway machines?
2021-03-10 06959, 2021
zas
I removed 20gb of unused images on paco, thanks to jenkins
2021-03-10 06908, 2021
ruaok
I keep getting kiki SMART emails.
2021-03-10 06934, 2021
sumedh joined the channel
2021-03-10 06939, 2021
zas
ruaok: yes, I know, that's not *that* urgent, in fact this alert is based on a counter, rather than a real state of the SSD. I want to deploy gitzconsul & serviceregistrator before messing with that
2021-03-10 06950, 2021
ruaok
ok
2021-03-10 06953, 2021
alastairp
zas: sorry about that. I tried to clean up some more images earlier in the week but didn't manage to do 20gb
2021-03-10 06907, 2021
ruaok
did we miss our deadline for consul upgrade again or is that coming soon now?
specifically, see the todo list that I added. It does it for one type of tests, but not the others. We should implement it in CB too
2021-03-10 06903, 2021
zas
we missed it, I'm waiting for feedback on gitzconsul & serviceregistrator, so if anyone wants to help testing...
2021-03-10 06957, 2021
zas
those are ready, but I wouldn't say "production-ready", mainly because I'm the sole that worked on this (bitmap did some testing a while ago though, and it helped)
2021-03-10 06945, 2021
zas
so if anyone can take some time this week, and runs local tests and report/comment, it'd be great
2021-03-10 06927, 2021
Lotheric has quit
2021-03-10 06951, 2021
zas
alastairp: I used the Hammer: docker system prune -a, dangerous but when a disk reaches 95%, violence is an option
2021-03-10 06950, 2021
ruaok
alastairp: fetching the Sentry DSN at build time means that the DSN will be included in publicly available images. Is that desired? acceptable?
ruaok: it won't be fetched in at build time. it will be passed as an environment variable in docker startup script
2021-03-10 06912, 2021
ruaok
ah. right. that's just fetching a binary. sorry.
2021-03-10 06920, 2021
ruaok
ugh. binary.
2021-03-10 06927, 2021
alastairp
ah yes, that's the app to submit messages
2021-03-10 06905, 2021
MajorLurker joined the channel
2021-03-10 06939, 2021
c1e0_ is now known as c1e0
2021-03-10 06902, 2021
MajorLurker has quit
2021-03-10 06913, 2021
atj
this KEYDOESNOTEXIST pattern seems like a bit of a hack
2021-03-10 06912, 2021
alastairp
it is
2021-03-10 06940, 2021
alastairp
the problem is, consul-template blocks if it can't find a key until that key appears, and gives no indication that this is what's happening
2021-03-10 06954, 2021
atj
is that not configurable?
2021-03-10 06923, 2021
alastairp
no
2021-03-10 06943, 2021
alastairp
well, kind of - the {{key}} template blocks, the {{keyOrDefault}} one doesn't :)
2021-03-10 06901, 2021
Rohan_Pillai joined the channel
2021-03-10 06908, 2021
atj
right, so you want to treat it as an error but consul doesn't allow it
2021-03-10 06935, 2021
alastairp
right, it would be nice to have an option to consul "quit with a status code if you're missing a key"
2021-03-10 06923, 2021
alastairp
but part of the benefit of consul is you can use it to dynamically rewrite files based on what data is available anyway - "If the primary database server is up, use x as config option, otherwise get it from the standby"
2021-03-10 06937, 2021
alastairp
so failing on a key not being there kind of defeats some of its benefits
2021-03-10 06958, 2021
alastairp
so we're kind of implementing it ourselves by writing the sentinel and then grepping for it
2021-03-10 06903, 2021
Darkloke has quit
2021-03-10 06948, 2021
alastairp
(we've had a number of cases in the last few years where we forget that this is the behaviour and we sit there watching docker logs wondering why nothing is starting up and our config file isn't being generated)
2021-03-10 06947, 2021
atj
I can imagine, that behaviour is not intuitive
2021-03-10 06903, 2021
Lotheric joined the channel
2021-03-10 06930, 2021
_lucifer
alastairp: i see. regarding image cleanup, how about we setup a cron job for it?
2021-03-10 06913, 2021
sampsyo has quit
2021-03-10 06924, 2021
alastairp
_lucifer: though, theoretically a test process should clean up after itself anyway
2021-03-10 06951, 2021
alastairp
there is some cleanup code in test.sh floating around, but it's buggy, which is why disk usage keeps increasing
2021-03-10 06956, 2021
_lucifer
alastairp: yeah, but some of these images might be reused in other tests, no?
2021-03-10 06912, 2021
alastairp
if it's still a problem after we apply this, we can look at a cron job
2021-03-10 06932, 2021
alastairp
nope - the way I did the cleanup is that we only remove images for finished tests
2021-03-10 06955, 2021
alastairp
but also, `docker image rm` will fail if a container using it is still running
2021-03-10 06902, 2021
alastairp
so it'll just get cleaned up the next time around
2021-03-10 06914, 2021
_lucifer
makes sense
2021-03-10 06928, 2021
alastairp
actually, maybe my process is overly complex (I say "delete all images made earlier to the one that I've just run")
2021-03-10 06937, 2021
alastairp
but images are tagged with the run number in jenkins
2021-03-10 06910, 2021
alastairp
oh - now I remember. I didn't want to just delete the image that we built for this test, because it means we'd have to run the build process from scratch on each test
2021-03-10 06917, 2021
alastairp
and it would make test runs even longer
2021-03-10 06917, 2021
sampsyo joined the channel
2021-03-10 06956, 2021
ShivamAwasthi joined the channel
2021-03-10 06916, 2021
_lucifer
that was my concern but i see you got it covered :D
2021-03-10 06905, 2021
atj
alastairp: possibly stupid suggestion, but could you define a MISSING_SERVICES list in config.py, then use {{ if keyExists "X" }} VARIABLE = {{ X }} {{ else }} MISSING_SERVICES.append("X") {{ end }}
2021-03-10 06927, 2021
atj
then check for MISSING_SERVICES in the application code and exit with a suitable message and code
2021-03-10 06911, 2021
c1e0_ joined the channel
2021-03-10 06926, 2021
c1e0 has quit
2021-03-10 06933, 2021
alastairp
atj: yeah, that's a possible option too
2021-03-10 06906, 2021
alastairp
although one benefit to the way that we've done it is that developers don't have to write a 4 line template for each variable
2021-03-10 06929, 2021
alastairp
(I guess we could abstract it out into a consul function)
2021-03-10 06952, 2021
alastairp
keep in mind also that we have config.sh rendered with consul too. the grep solution works on that out of the box too :)
2021-03-10 06937, 2021
_lucifer
i don't think we can a consul function without creating a plugin
2021-03-10 06947, 2021
_lucifer
*can create
2021-03-10 06910, 2021
alastairp
no, you can do it in the templating language. musicbrainz has something like that
2021-03-10 06916, 2021
c1e0_ is now known as c1e0
2021-03-10 06926, 2021
_lucifer
oh! interesting. i was looking for that in consul docs but didn't find it.
for shell snippets you could do the same thing, and have a function to check the shell variable
2021-03-10 06902, 2021
atj
the grep solution as it seems quite brittle
2021-03-10 06934, 2021
atj
*I was thinking of how to replace
2021-03-10 06956, 2021
_lucifer
alastairp: i think this could be used for the rabbitmq thing we had in LB as well.
2021-03-10 06920, 2021
_lucifer
{{ $var := executeTemplate "custom" }}
2021-03-10 06947, 2021
_lucifer
as this allows us to store the template output in a variable
2021-03-10 06946, 2021
alastairp
atj: yeah, I see the value in doing it like that
2021-03-10 06947, 2021
ShivamAwasthi has quit
2021-03-10 06959, 2021
alastairp
out of curiosity, why do you think that grep is brittle in this case?
2021-03-10 06933, 2021
alastairp
I'm not sure I can think of any cases where it might causes problems
2021-03-10 06941, 2021
c1e0_ joined the channel
2021-03-10 06902, 2021
shivam-kapila
( alastairp: hi. I am here. Ping me once you are free)
2021-03-10 06901, 2021
atj
Perhaps brittle is the wrong word. I think that using magic values is generally a bad idea, because you're using it to communicate state without any context.
I'm sure it'll work fine for some time, and then someone will change the way the config file is rendered and break it because they don't understand the context
2021-03-10 06947, 2021
alastairp
right, good point
2021-03-10 06957, 2021
atj
not that my suggestion is perfect in that regard
2021-03-10 06909, 2021
alastairp
if we use app.config.from_pyfile, it imports it, but that means we'd have to do something like catch a custom exception here
2021-03-10 06924, 2021
alastairp
which again means that you have to remember to catch it if you import the config file somewhere else
2021-03-10 06942, 2021
alastairp
I kind of get the feeling that all solutions have problems, and I went for the easiest to implement...
2021-03-10 06944, 2021
atj
yeah, I don't think there's a great solution
2021-03-10 06953, 2021
atj
yes, totally agree
2021-03-10 06958, 2021
alastairp
shivam-kapila: sure, I'm about to head off to lunch. will you be around in an hour?