Has anyone else seen increased memory usage of their instance after upgrading to 2.9?

I'm currently noticing a more or less linear growth in memory usage (with a small bump down each night) ever since I installed 2.9.2

Yesterday, I had to disable the 6GB memory limit on the lxd container I keep my Mastodon in, as there was just too much swapping going on.

actually higher memory usage here too. running proxmox vm with 4gb actually but think will enhance to 6gb

@galaxis My instance is using 0,98GB and the usage is always more or less the same.

@nipos Naja, meine Instanz läuft schon seit über zwei Jahren, und allein die Postgres-DB ist knapp 12GB groß... Hab' außerdem auch Elasticsearch am laufen.

@galaxis Ja gut,da kann man schon mal ein bisschen RAM verbrauchen.Aber solche sprunghaften Anstiege konnte ich bei mir eben nicht feststellen.

@nipos Ich werd' mal schauen müssen, welcher Prozeß dafür verantwortlich ist. Dachte halt, jemand anderes hätte das evtl. auch beobachtet und schon untersucht - mal abwarten, ob es über Nacht noch Rückmeldungen gibt.

@galaxis Yes, my smallish instance is currently using 14GB 8GB of which is from sidekicks workers. This is a sizable increase I believe.

@debugninja @galaxis thanks for the heads up! Will wait for a fix before upgrading

If you want a fix then you have to open an issue on GitHub 😉
@debugninja @galaxis
My RAM-Usage is the same during the last three months but I use fully dockerized Mastodon-Instances...

@markusblogde @emanuel @debugninja No use opening an issue when I don't even know if it's a real problem.
I'll observe this for a couple more days.

@emanuel @galaxis I pulled some historical data and its a slight increase but not a large one compared to normal for me so I don't know if its 2.9 specific.

@debugninja Yeah, still going on here, just mitigated by giving out more memory. Biggest chunk is sidekiq for me too.

@debugninja Hrm. I also notice that I upgraded on Jun 22, and from that day I got quite a spike in failed sidekiq requests. Not sure what's up with that (but also out of time for debugging tonight).

Wonder if there's any connection between the many failed sidekiq jobs on my Mastodon instance and this commit ("Fix HTTP requests to IPv6 hosts"):

This seems to fix an error introduced before the release of v2.9.0 in this change:

While I only publish an IPv4 address for my instance right now, it does have IPv6 connectivity (and uses v6 for outbound connections).

Guess I'll try updating to master.

Hrm. Doesn't really look any better on that front.

To the contrary - running out of memory more quickly now...

Sign in to participate in the conversation
INFRa Mastodon

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!