#metabrainz

/

      • zas
        if someone deployed something that may use the db on sunday, please tell ;,)
      • bukwurm
      • zas
        apart that, i tried to setup an alert for packet replication issues, but i need to start over, nagios was reluctant to help me
      • bukwurm
        CatQuest: There
      • CatQuest
        πŸ™‡ thank you bukwurm
      • zas
        plus usual PRs reviews, edits, etc...
      • fin. ruaok ?
      • ruaok
        k
      • normal week for me.
      • accountants, a meeting with the lawyers to cleanup a mess that c created.
      • bank account still not unfrozen, starting to doubt it will ever be.
      • bukwurm
        CatQuest: It's not, I created the gif online, conversion made the video slow.
      • ruaok
        helped with replication packet issues, realized that our docs are woefully out of date. we need to work on that.
      • did some tiny metabrainz navbar cleanup and deployed a link to our shop which promptly generated one sale.
      • thanks Major_Lurker!
      • CatQuest
        :D
      • ruaok
        I'm going to be AFK wednesday afternoon - sunday.
      • if you don't get me your invoice by tomorrow afternoon, you'll need to wait until monday.
      • fin.
      • iliekcomputers: go
      • bukwurm
        CatQuest: Oh sorry, I thought you were asking why it's slow πŸ˜…
      • CatQuest
        nah :D
      • iliekcomputers
        I'll go in a sec, I'm deploying CB rn
      • Freso
        (iliekcomputers is last one left on the list. If anyone else wants to go, now's the last time to speak up.)
      • bukwurm
        It's WIP, not live. Let's talk about it after the meeting.
      • iliekcomputers
        oh
      • okay
      • CatQuest
        Freso: hey!
      • iliekcomputers
        Can i go after next topic then?
      • Freso
        Oh, right. CatQuest is left too! Sorry!
      • CatQuest: Go!
      • CatQuest
        bukwurm: yup
      • iliekcomputers is next afaik?
      • iliekcomputers
        CatQuest: go, I'll go next
      • gcilou joined the channel
      • CatQuest
        πŸ‘
      • just been working a little bit on instruments and releases.
      • had a very interesting discussion with LordSputnik and bukwurm
      • in which bukwurm made a document summarising what we talked about
      • (and has been working on it too ass is the review he made :D)
      • she?
      • but mostly preparing seeds and stuff for spring took most of my time this week, should be back on form next week.
      • it's valborgsnatten this night so I'll be celebrating >:D
      • 🧹
      • Finn!
      • iliekcomputers go!
      • iliekcomputers
        ok
      • this week was much more productive than the last few, I'll do updates project wise
      • Freso
        (iliekcomputers is really the last one left on the list now. :) If anyone else wants to go, *now*'s the time to speak up.)
      • iliekcomputers
        LB: worked on moving data to the new BQ schema, ruaok pointed out an interesting problem, we discussed solutions to that, gonna code up something today.
      • CB: worked on comments on reviews, and a bunch of open PRs. proud that the only PRs left now are schema changes. thanks ferbncode!
      • ferbncode
        \o/
      • chhavi_
        yayy. Param on fire. Much parties this week? :D
      • CatQuest
        \o/
      • iliekcomputers
        AB: met with rsh7 and ferbncode over lunch to discuss how to approach her project in the best way, this is a first for me as a mentor. (was kinda productive)
      • chhavi_: indeed. :)
      • rn, working on changes to CB to reduce load on postgres.
      • fin!
      • CatQuest
        !m iliekcomputers
      • BrainzBot
        You're doing good work, iliekcomputers!
      • Freso
        Alright! Moving on!
      • zas: pg spof & scalability
      • zas
        basically we need to improve pg stuff
      • currently we have bowie (master postgresql) and queen (slave, underused)
      • while it provides some redundancy, we have any load balancing yet
      • iliekcomputers
        moving CB (beta and prod) to queen, although I'm not sure how much load CB causes anyways.
      • zas
        so that's a single point of failure (spof): if bowie goes down, we have no automatic switchover
      • CatQuest approvesapproves of using Queen more
      • CatQuest
        :D
      • zas
        discussing possible solutions is too long here and now, but i'd like people with ideas to tell, we'll try to address those problems asap
      • bitmap
        some interesting comments about pgpool there https://www.reddit.com/r/PostgreSQL/comments/6k...
      • zas
        our lack of experience in this field is the core issue, if anyone knows someone that could help on this stuff, please make the link ;)
      • bitmap
        but I think pgpool or looking closer at citus are the best options I see at present...
      • zas
        for sure we have to dig those options
      • but from what i read none are exactly easy to set up
      • bitmap
        certainly not pgpool
      • zas
        that's why i think we need to find people willing to help that have experience with whichever tool we go for
      • we can start by postgresql channels
      • ruaok
        do a blog post about it. asking for recommendations.
      • that has worked really well in the past.
      • CatQuest
        it has
      • <3 our community
      • zas
        yup i agree
      • Freso
        Anything else you need to discuss about this?
      • zas: ?
      • zas
        not really, but that's something we need to put in priority before it becomes a very serious issue, the db is the core of our ecosystem...
      • Freso
        Absolutely.
      • zas
        and regarding the current situation i'd say it's urgent ;)
      • Freso
        So: make a blog post as well as reach out to people in pgsql channels?
      • CatQuest
        πŸ‘
      • Freso
        zas: Not saying it isn't urgent, just wondering whether it's time to close up the meeting. :)
      • zas
        it is, i'll do the blog post during this week
      • Freso
        Alright.
      • Thanks for your time everyone!
      • </BANG>
      • zas
        bitmap, yvanzo: any idea about what started yesterday around 16:00 UTC (which is the start of the load issue) ? i looked at many things and still cannot explain the load on bowie. We know restart/reboot changed nothing in this regard and it isn't due to higher than normal number of requests.
      • yvanzo
        Leo_Verto: Let me know what resources you need for SpamBrainz.
      • Leo_Verto
        I'd really like to get my hands on some sort of preliminary spam data set
      • yvanzo
        zas: me neither
      • Leo_Verto: Ok, do you have an MBS setup at hand?
      • Leo_Verto
        run some basic data analysis on it, get a feel for what exactly we're dealing with here
      • yvanzo: nope, I'll get one set up asap
      • bukwurm
        CatQuest: I am planning on having integrated search include all types of entities, but raise warning only when a same entity exists with same name (and then making addition of disambiguation compulsory).
      • zas
        yvanzo: did you restart all mb website & ws containers ?
      • yvanzo
        bitmap, zas: it doesn’t seem to be related to latest MB release either.
      • zas
        there are extra disk activity (especially swap) on many nodes when it started : https://stats.metabrainz.org/d/000000052/hetzne...
      • yvanzo
        zas: on Saturday
      • zas
        but you didn't today ?
      • bukwurm
        CatQuest: I plan on modifying the name matching function to add links to entities with same name, exactly as it is in MB.
      • yvanzo
        no
      • bitmap
        nothing cron-related or deployment-related I'm aware of
      • yvanzo
        zas: I restarted beta containers
      • (today)
      • zas
        ok, let's restart all prod containers to be sure
      • CatQuest
        bukwurm: that does sound good
      • (and very inline with what mb currently does)
      • zas
        bitmap: can you do it ?
      • bukwurm
        And maybe make the warning disappear if the user enters a disambiguation?
      • bitmap
        all mbs prod containers?
      • zas
        yes, website & ws
      • bitmap
        ok
      • bukwurm
        CatQuest: Of course, it'll all be inline.
      • CatQuest
        yep, since then it indicates "yes this was the problem, you solved it no more a prolem"
      • yes!
      • chhavi_
        bukwurm: yep. That should be. I
      • CatQuest: is keeping track of you, so I need not be worried. Couldn't help but be curious. :) :p
      • iliekcomputers
        ok, CB prod now using postgres-slave
      • CatQuest
        I still think that the popus of the other bit should be inline and expandalbe sections . but LordSputnik said not "for now"
      • iliekcomputers
        now, CB beta.
      • bukwurm
        Making warning disappear - that'll be 2 minute job max. Let me do it right now.
      • chhavi_: All views are more than welcome.
      • :D
      • CatQuest
        one thing to keep i nmind as that mb is having a redesign. we should eventually work on getting parity between bb and mb apearance wize
      • ruaok
        iliekcomputers: thanks!
      • CatQuest
        take the best of both worlds so to speak
      • bukwurm, chhavi_ definitly
      • yvanzo
        zas, bitmap: restarting mb prod containers
      • github joined the channel
      • github
        [critiquebrainz] paramsingh opened pull request #207: Merge production into master (master...merge-branch) https://git.io/vpRQf
      • github has left the channel
      • zas
      • iliekcomputers
        ruaok: approve and I'll deploy?
      • bukwurm
        CatQuest: I think as long as functionality remains same, it's all good no matter how it looks.
      • CatQuest
        I'm working for man "user" prespective. a "design perspective" is chhavi's and als oimportant
      • bitmap
        yvanzo: ok, I just did bootsy and burnside
      • you can do the rest if you'd like :)
      • zas
        ok, we had the memory issue on ws containers at the time it started
      • CatQuest
        bukwurm: the thing with that is that look does finger into functionality. somewhat
      • zas
        it doesn't explain why it didn't stop after it came back to normal though
      • CatQuest
        it needs to "look" i na way that makes it easy to use ()ie funtion is obvious