autoscaling
autoscaling copied to clipboard
neondatabase
Postgres vertical autoscaling in k8s
## Environment make k3d-setup ## Steps to reproduce ``` [ 0.392350] Run /neonvm/bin/init as init process [ 0.401884] ext3: Unknown parameter 'mode' [ 0.402145] ext2: Unknown parameter 'mode' [ 0.402356]...
WIP- thinking of this because of #827. Maybe we want to actually unify the reconciler types?
``` {"level":"panic","ts":1699122550.6897972,"logger":"autoscale-scheduler","caller":"plugin/plugin.go:236","msg":"deadlock checker for AutoscaleEnforcer.state.lock panicked","error":"likely deadlock detected, could not get lock after 1s","stacktrace":"github.com/neondatabase/autoscaling/pkg/plugin.makeAutoscaleEnforcerPlugin.func13.1\n\t/workspace/pkg/plugin/plugin.go:236\nruntime.gopanic\n\t/usr/local/go/src/runtime/panic.go:884\ngithub.com/neondatabase/autoscaling/pkg/util.(*ChanMutex).DeadlockChecker.func1\n\t/workspace/pkg/util/chanmutex.go:102\ngithub.com/neondatabase/autoscaling/pkg/plugin.makeAutoscaleEnforcerPlugin.func13\n\t/workspace/pkg/plugin/plugin.go:240"} ``` Encountered during a 10k VM cap test https://github.com/neondatabase/cloud/issues/7772
## Environment everywhere ## Steps to reproduce To make it more explicit add `time.Sleep(10 * time. Second)` to `neonvm/runner/main.go` somewhere before executing the QEMU command. ## Actual result NeonVM controller...
## Problem description / Motivation Currently we use a single gauge for the "autoscaling stuck" metric exposed by the autoscaler-agent. This is nice because it's simple. However, a downside of...
## Problem description / Motivation Currently it's difficult to see "typical" billing request duration timings. Fixing the issue with `lastSendDuration` (ref #787) _should_ help somewhat here, but still doesn't show...
Original title: "vhost backends exhausted" ## Steps to reproduce related to large VMs ## Expected result ## Actual result ## Environment ## Logs, links -
## Problem description / Motivation There's a few reasons why we might want to switch to using systemd: 1. Better lifecycle management for "daemons", which may help with #577 2....
Fixes #680, see there for detail on motivation. tl;dr: this fixes a known category of bugs, and AFAICT is a pre-requisite for using the VM spec as a source of...
Follow-up change to #371 that I'd wanted to make at the time, but focused instead on finishing it, because this is really just a refactoring. Marked WIP until full cleanup...