sean/olly
Watch
1
0
Fork
You've already forked olly
 0
Code Issues Pull requests Projects Releases 60 Packages Wiki Activity

fix(metrics): scope the seed to one named replica #45

Merged
binjovi-bot merged 1 commit from chore/vm-restore-fix into trunk 2026-09-03 20:58:16 +00:00
Conversation 0 Commits 1 Files changed 1 +27 -4
sean commented 2026-09-03 20:57:45 +00:00
Owner
Copy link

The first guard tested for /storage/data, which vmsingle creates on an empty volume, so the seed skipped itself on the very replica it was written for.

Size is no better: victoriametrics-1 had already collected 1.3 GB from vmagent in the hour since its volume was replaced, against about 19 GB on a populated replica, and it keeps growing.

Guard on the pod name instead. SEED_POD names the one replica to seed and every other pod exits immediately, so this can never overwrite a replica holding the only good copy. A sentinel file makes the seeded pod idempotent across restarts.

vmrestore wants an empty target, so the seeded replicas directory is cleared first. Nothing unique is lost: vmagent dual-writes every sample to both replicas, so whatever this replica collected is a strict subset of the source.

A fresh 19 GB backup of replica 0 was taken at 20:56 so the seed is current rather than four hours stale.

https://claude.ai/code/session_01MdSbMhzabSbpG8TtP9Ur3H

The first guard tested for `/storage/data`, which vmsingle creates on an empty volume, so the seed skipped itself on the very replica it was written for. Size is no better: `victoriametrics-1` had already collected 1.3 GB from vmagent in the hour since its volume was replaced, against about 19 GB on a populated replica, and it keeps growing. Guard on the pod name instead. `SEED_POD` names the one replica to seed and every other pod exits immediately, so this can never overwrite a replica holding the only good copy. A sentinel file makes the seeded pod idempotent across restarts. vmrestore wants an empty target, so the seeded replicas directory is cleared first. Nothing unique is lost: vmagent dual-writes every sample to both replicas, so whatever this replica collected is a strict subset of the source. A fresh 19 GB backup of replica 0 was taken at 20:56 so the seed is current rather than four hours stale. https://claude.ai/code/session_01MdSbMhzabSbpG8TtP9Ur3H
fix(metrics): scope the seed to one named replica
All checks were successful
binjovi/ci Binjovi completed the frozen plan
Details
399f08c27d
 
The first guard tested for /storage/data, which vmsingle creates on an
empty volume, so the seed skipped itself on the very replica it was
written for.

Size is no better: victoriametrics-1 had already collected 1.3 GB from
vmagent in the hour since its volume was replaced, against about 19 GB on
a populated replica, and it keeps growing.

Guard on the pod name instead. SEED_POD names the one replica to seed and
every other pod exits immediately, so this can never overwrite a replica
holding the only good copy. A sentinel file makes the seeded pod
idempotent across restarts.

vmrestore wants an empty target, so the seeded replica's directory is
cleared first. Nothing unique is lost: vmagent dual-writes every sample to
both replicas, so whatever this replica collected is a strict subset of
the source.

Claude-Session: https://claude.ai/code/session_01MdSbMhzabSbpG8TtP9Ur3H
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
sean/olly!45
No description provided.