home.social

#colima — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #colima, aggregated by home.social.

fetched live
  1. Ustedes, gobierno, quitan carteles de nuestra gente desaparecida. Nosotros, ciudadanos, quitamos carteles de sus campañas anticipadas. #Colima . No somos iguales.

  2. about to attempt to bring my #SelfHost back up. It took over an hour for #newOldiMac to compile and install all the requirements for the qemu emulator for colima so that's why it's been a minute. Ended up taking all day lol.

    Anyway, it's running now. I'm able to run a colima context under either the original vz or the new qemu profile.

    If I get the db and redis running on it successfully, we'll see if it's more stable as hoped!

    #mastoAdmin #Colima #macOS

  3. “what are you trying to do now?”
    you might be asking.

    Colima on macOS Ventura and later uses the Apple-supplied 'vz’ engine.

    #newOldiMac is running Ventura.

    So maybe docker was interacting poorly with that engine, especially because it was the first macOS version with it… maybe it's buggy?

    So I'm installing ‘qemu', which is the native colima engine on all pre-Ventura Macs and Linux.

    building it is taking a bit…

    #selfhost #footiMac #newOldiMac #mastoAdmin #qemu #colima #cantHelpMyself

  4. With all the drama today fixing Colima and docker on #newOldiMac and then the upgrade to v4.6, who knows what could have happened, but it all worked out!

    Colima and Docker *seem* happy with the changes. All the docker containers have their own settings. The webcams are pinned to one core, and the db and redis are allowed to use all. Memory is customized for each.

    So far, no crashes! Fingers crossed it is up in the morning!
    #mastodon #mastoadmin #docker #colima

  5. Alright well, no luck on finding the command used to invoke a container. But I’ve tweaked the containers nonetheless that I have running in #colima on #newOldimac. It *feels* like docker is overshooting what colima is able to give it or one container is stepping on another's toes.

    So I have ‘pinned' the three webcam containers to one CPU core and allowed the mastodon db and redis containers to continue to access all 4 cores.

    I also trimmed down memory available for the webcam containers to only what they need. 256MB normal reservation shrinking to 128MB if the other two containers need a little extra.

    I have yet to see the DB container demand more than 2GB of memory, so 4GB should be plenty.

    I will probably trim down the memory for the redis container as well as it has never gone over about 128MB. So I can probably cap it to 1GB with a reservation to 512MB.

    I enjoy the tweaking…

    docker stats should be more verbose. I enjoy the verbose.
    #docker #imac #colima

  6. I just reconfigured the Colima VM on #newOldiMac to try to get a handle on the instability.
    I love doing stuff in a command prompt on an iPad.
    It’s cool.
    (Terminal# app on iPad)
    #docker #colima #imac #ipad #ipadOS

  7. I think it is funny, ironic, nerdy, and meta (the proper literal sense) that I have maxed out the 2017 iMac by wanting to install btop while watching it be maxed out on htop.

    And yet... it's still posting this, so if stuff doesn't break down now, surely it never will? Even the swap is having a moment. Ooof. I need to be nicer to my machines. lol

    #newOldimac #selfhost #mastoAdmin #footiMac #docker #colima #homebrew

  8. just tweaked/upped web concurrrency and max threads settings in my .env.production settings for Mastodon to give #footiMac a few more processes to handle requests and things are humming along nicely.

    I think a big takeway for me on this #selfhost journey is that except on the very smallest one person and minimal followers instance, the database (and possibly redis) function should always be on a different machine or container from mastodon.

    #footiMac was clearly having real trouble handling the database activity once it got busy. Even with an SSD now with two bad blocks (which may have been caused by that over-activity), it is way more stable.

    #newOldiMac #docker #mastoAdmin #colima #iMac

  9. It's a day later and #footiMac and #newOldiMac are still up and running. Disk usage is stable. Current RAM usage for the database has gone up to 3GB from 1GB since 9PM last night. I'd like to know why. It's not any busier.

    I'm running the backup script then later today I'll see if I can investigate more on why the RAM usage fluctuates so much.

    I would like to know why it is more stable. Is it because I've simply given it enough room to expand and contract in the colima VM?

    Is it because I set the clocks to the same timezone?

    Inquiring minds... but I am happy that it is happy. And I have learned a lot and gotten more comfortable with docker. Which has been a goal almost since I joined Mastodon in Nov 2022.

    I’m hoping to look for a new macMini now to replace #newOldiMac. If I can get that an SSD to replace the failing SSD in #footiMac and a large HDD for backup that'd be great.

    #docker #mastoAdmin #colima #iMac

  10. success.
    docker restart does not jeopardize anything as long as a volume is specified at runtime to preserve the data.

    successfully restarted with new conf file and we're now green across the board and hopefully tuned for the slightly larger resources available from #newOldiMac

    #docker #mastoAdmin #docker #colima #postgresql #selfhost

  11. Edit: question answered successfully! Ty!
    OK folks, I need some #docker advice for those gurus out there. A boost would be appreciated to get it in front of knowledgeable eyes.

    I have a container running postgresql for mastodon. I would like to tweak the conf file for postgresql now that I have things going.

    What is the safest way to do this on a running container?

    Can I stop the instance with docker stop and then restart it and it will simply see the new postgresql.conf specified in the original docker run command?

    I'm getting more comfortable with docker.. but not *that* comfortable, so I'm still worried that if I restart it'll break somehow lol.

    I want to make sure I do this right so I don't wreck what appears to be a running and reasonably happy container :)

    I'm running docker in a colima virtual machine on macOS.

    #mastoAdmin #docker #colima #postgresql

  12. ok. docker backup script is solid.
    it will:
    1) mastodon db dumped in container and copied outside
    2) redis file saved in and outside
    3) both containers snapshotted and saved outside.
    #selfhost #mastoadmin #colima #macos #docker #jinx

  13. this is the longest that my #selfhost instance has been online in 5 days lol. Starting to feel confident. Probably jinxing it…

    #mastoadmin #colima #macos #docker #jinx

  14. making a script to backup the postgresql and redis docker containers... so I can have this run everyday to ensure I don't lose anything if/when docker collapses.

    I also noticed that the docker containers all use UTC time, which is annoying, and who knows maybe that has something to do with the instability? so i synced it to America/Vancouver

    Also... I know "America" is the continent, not the country in this context. But there is more than one Vancouver on the American continent. Canada/Vancouver would be better, no?

    #random #selfhost #mastoadmin #docker #colima

  15. ok, I've managed to clear 250GB of space on my iMac and give colima access to 200GB.

    Now that I have a handle on how to monitor disk space both in colima and in each docker container, I'll be monitoring it closely.

    The bad news is I had to revert back to a database from Friday night so I've lost some content. But I'm again increasing my understanding of working within a docker environment within a colima virtualization.

    I’m updating the docker images for both the db and the redis now and will take current backups from both so I can be closer to realtime.

    I'll get there! I refuse to let this beat me. I feel like I'm close!

    Kind of in the background of all the drama on #newOldiMac trying to get this colima and docker stuff to behave, trusty #footiMac is much happier now that some load is off. But I still have my eye on a new Samsung branded and hopefully reliable 2TB SSD to replace the one that is failing that gave rise to all of this.

    screenshot of realtime “hacking" according to @Jade lol.

    #mastoAdmin #selfhost #colima #macos #linux #docker #footiMac #newOldiMac

  16. ok.. not sure how... but the database went away on #newOldiMac so now I'm restoring it again from Saturday night, so I'll lose my posts from Sunday. The tables were gone as if it had never been created... was it when I restarted the docker container and it started from a version with no database!?)

    But I have a volume set up so it should keep that the data regardless... but it didn’t?

    computers only do what they tell them to, so I'm sure this is a me-problem at some point :)
    #docker #colima

  17. So at least part of the errors I was seeing were erroneous.
    Turns out ivory caches something with notifications. After I went on each of my devices and found the clear cache for my account in Ivory, it is working fine.

    So really it is just the periodic loss of connection to the redis server. But at least it is self-restarting?

    #mastoadmin #redis #colima #macos #debian

  18. Ok, current theory is that redis is running up against a “vm.overcommit_memory” issue with the colima virtual machine. So it essentially runs out of memory no matter how much memory you give it.

    So I've started colima with a provision to set overcommit to 1 which ignores the overcommit (which seems bad, but ok)

    We'll see if that stabilizes things.

    #colima #macOS #docker #mastoAdmin

  19. yes!
    successfully restored!

    now to modify that colima/docker backup procedure so I don't forget the data next time... lol

    #mastoAdmin #selfhost #footimac #newOldiMac #colima #docker #learning

  20. oh.
    rookie mistake.
    i thought the docker container in colima would include the persistent data?

    Apparently.... not.
    thankfully I still have the original dump from #footiMac
    #colima #docker

  21. I appear to have pooched the colima instance a little. So I am going to take a snapshot of the docker containers and re-install colima with some better optimizations

    I gave it too much space I think. It's also got a bunch of useless docker images in there taking up space.

    #colima #docker

  22. looks like I am running out of space for #colima and its docker instances. This is forcing it to shutdown.

    Anyone familiar with how it or #docker allocates resources/disk space?

    I have one redis and one postgresql container.