neondatabase / autoscaling

Postgres vertical autoscaling in k8s

neondatabase/autoscaling Issues

Make metrics LFC number of hits, number of misses and working set size available to autoscaling algorithm
Updated 15 days ago10
autoscale-scheduler: Reduce log verbosity
Updated 16 days ago
Docker rate limiting causes e2e-tests to fail
Closed 23 days ago
Internal feature: Stateless scheduler
Updated 25 days ago
Docker rate limiting causes CI to fail
Updated a month ago
Gradually increase LFC quota without taking working set into consideration
Updated a month ago4
Bug: neonvm-controller spams new pod names when VM fails and RestartPolicy = Never
Closed a month ago
Bug: neonvm-controller (still) has log sampling enabled
Closed a month ago
Non-functional codebase improvements
Updated a month ago
Metrics and clearer logs for denied scaling
Updated a month ago
Bug: /neonvm/bin/resize-swap: command not found
Closed a month ago1
Add a clear indication of why resize-swap fails
Updated a month ago
neonvm: separate maps / counts for failure vs conflict
Closed a month ago1
e2e test 'restart-counted' is flaky
Updated 2 months ago
Expose current version and show it on dashboards
Updated 2 months ago
Bug: compute_ctl can't open neon.tech.log.0 after restart
Closed 2 months ago
0/3 nodes are available: 1 node(s) had untolerated taint {node-role.kubernetes.io/control-plane: }, 2 Insufficient neonvm/kvm.
Closed 2 months ago5
neonvm-controller reconcile failure metrics should differentiate k8s conflicts
Closed 2 months ago
neonvm-controller logs don't show reconcile duration
Closed 2 months ago3
Remove scheduler plugin "buffer" resources
Updated 2 months ago4
Bug: scp into guest-vm doesn't work
Updated 2 months ago1
Bug: stop treating downscale denies as an error
Closed 2 months ago
Alpine base images should be updated
Updated 2 months ago2
scheduler plugin should have metrics for time events spend waiting in the queue
Closed 3 months ago2
Epic: Switch to new scaling strategy
Closed 3 months ago1
Bug: `make manifests` fails with Go >1.21
Closed 3 months ago
Make sure that failing VMs is tested properly
Updated 3 months ago1
neonvm-vxlan.yaml and neonvm-multus.yaml files are missing
Closed 3 months ago2
Support resizeable swap
Closed 3 months ago1
Change plugin's `Reserve()` method so it cannot fail
Closed 4 months ago1
Get rid of VM watcher in plugin
Updated 4 months ago1
neonvm-controller does not update VM status if reconcile fails
Updated 4 months ago
Internal feature: refined compatibility matrix
Updated 4 months ago
Epic: automated autoscaling release workflow
Updated 4 months ago
Epic: Support fixed bigger compute units
Updated 4 months ago1
Bug: scheduler's critically delayed event processing has positive feedback loop
Closed 4 months ago3
scheduler plugin's failed scheduling metrics should include preferred AZ of the pod, if available
Closed 4 months ago
VM info should be available on runner pod, so scheduler doesn't need to watch VMs
Closed 4 months ago
Complete cold start dashboard
Closed 4 months ago2
Bug: scheduler has negative "buffer" value
Updated 4 months ago1
Epic: Support ARM AWS instances
Updated 4 months ago
Documentation for releases
Closed 4 months ago1
Epic: Solving the “pgvector 0.6” problem
Closed 4 months ago2
scheduler plugin should de-prioritize newer nodes
Updated 4 months ago
linter for unused function return value
Updated 4 months ago1
Scheduler leader election
Updated 5 months ago
Epic: Standard interface for running containers AND/OR systemd services in VMs
Updated 5 months ago
Epic: Productionize automated live migration
Updated 5 months ago
Epic: Productionize manual live migrations
Updated 5 months ago
Bug: unknown parameter 'mode'
Updated 5 months ago