upgraded to 1.5.1
presented without comment
more advances from deepmind, they seem to be speeding up
prevention steps for the future: added an additional node to the cluster, upped the replica count for sidekiq and streaming to 2 for better redundancy against single node failure. I have a generic monitoring/alerting solution written out, but haven't yet deployed it on this cluster, oops! Will get that deployed so I get better mobile downtime alerts on triggers like frequent pod restarts, 0 available replicas in a set, or no endpoints for a critical service.
and we're alive again. Finally caught what I believe was thrashing the nodes: "find /mastodon -path /mastodon/public/system -prune -o -not -user mastodon -not -group mastodon -print0 | xargs -0 chown -f mastodon:mastodon" in the entrypoint script.
honestly people trying to sell usernames on mastodon instances will never not be funny
Alright this instance is now up to date on 1.4.6
I'll continue updating as seems prudent, still noticing lots of goofy QC issues in masto generally so we'll likely stay behind the bleeding edge intentionally to continue giving bugs time to shake out before I update.
@sull @sovereign_bleak Screencast of the update process/kube tour is online finally. I was putting this off until I had more time to polish and edit a recording, but it doesn't look like that time is going to materialize anytime soon so a rough cut is all I can give for now: https://www.youtube.com/watch?v=KEEgTc63NuU
gonna give 1.3.1 a day or so to shake out on other instances and then will probably upgrade here. curious to read more about what the privacy complaints are about but no time at the moment
anyone who wants to call what we're in "late capitalism" is in for a surprise in a few years
alright I'm assuming there must be some kind of model inspection on startup and my sidekiq instances came online before the db migration completed. Killing their pods seems to have resolved my previous issue.
Also seeing some errors fetching remote status updates, looking into it now. Related to newly added language field on the status model. The field was added during a schema migration and has been verified to exist on the database, but ActiveModel seems to think otherwise.
This update moved a bunch of important URLs without giving a redirect for all the existing links out there 🤔
doing updates on this instance in the next hour, shouldn't have any downtime.
Toying with the idea of writing more about mastodon sustainability. Boost/Star if you would read about instance fragility, content duplication, user/account migration between instances, and just the general adaptation processes this system has today/might need soon.
An Empirical Analysis of Linkability in the Monero Blockchain
Apologies to everyone I had to leave out! (your instances are great too)
From Witches to Dolphins, These Are the Communities That Make Mastodon Great https://motherboard.vice.com/en_us/article/from-witches-to-dolphins-these-are-the-communities-that-make-mastodon-great #eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
putting out fires and plunging toilets for this mastodon instance - just a dude in oakland
email at email@example.com
capitalism.party is a paid signup Mastodon instance funded directly by users purchasing accounts for just $5. An inexpensive alternative to free signup platforms, we impose direct economic cost on trolls who want to avoid blocks by creating many accounts. This instance will actively respond to any problematic users.