mara/hyperhive
mara/hyperhive
Watch
0
0
Fork
You've already forked hyperhive
 0
Code Activity

otel: collectors drop their own export-failure lines instead of shipping them

Browse source
This commit is contained in:
atlas 2026-11-14 22:13:39 +01:00 • committed by mara
commit 3fa926d7d6
1 changed files with 115 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

11
docs/4-implementation/swarm/telemetry.md
View file

@ -309,6 +309,17 @@ every other exporter still arrive. The container mounts the host's journal
read-only at `/run/host-journal`, not over its own
`/var/log/journal`.
Every collector that reads a journal drops its own export failures before it
exports: the hive collector, the agent and service-container forwarders, and
the swarm collector's host-journal pipeline. A `filter` processor
(`nix/collector-own-export-failures.nix`) matches records from the unit
`opentelemetry-collector.service` whose message contains `Exporting failed` or
`sending queue is full`. Without it, an ingest outage makes each collector log
one line per failed batch, its receiver reads the line back and pushes it, and
the pushes add to the outage. The collector's startup, configuration and
receiver errors still ship. The failure itself stays visible in the collector's
`otelcol_exporter_send_failed_*` metrics.
**Both store exporters are unconditional**, and `deploy.victoriametrics.enable`
doesn't gate them: that option says this host _runs_ the store, while the swarm
has one either way, reached by its swarm name through the gateway.