Skip to content

fix(two-node): run the event collector as a python3 module, not the deleted Go binary - #956

Draft
mxsrc wants to merge 1 commit into
mainfrom
fix-two-node-event-collector-command
Draft

mxsrc wants to merge 1 commit into
mainfrom
fix-two-node-event-collector-command

Conversation

@mxsrc

@mxsrc mxsrc commented Nov 20, 2026

Copy link
Copy Markdown
Contributor

Summary

  • sbcli's translate-go rewrote the two-node event collector from a Go binary (tools/event-collector, installed at /usr/local/bin/sb-event-collector) into simplyblock_core/services/two_node_event_collector.py, and removed the Dockerfile stage that built the Go binary.
  • managementapi.go's monitoringServices() still pointed the two-node-event-collector container at the deleted binary path, so any control-plane image built from translate-go (or once it merges, from sbcli main) crash-loops that container: stat /usr/local/bin/sb-event-collector: no such file or directory.
  • Confirmed live on cluster91 deploying a two-node test cluster with the translate-go image.

Changes

  • two-node-event-collector now uses module: "simplyblock_core/services/two_node_event_collector.py", same as every other monitoring service.
  • Dropped the now-dead twoNodeEventCollectorBinary const.
  • Updated the two tests that pinned the old binary path (TestTheServicePoolsRunWhatTheyDeclare's special-case branch, TestTheTwoNodeServicesNameWhatTheImageShips's want map).

Test plan

  • make test (full suite, envtest included): all packages pass, controlplane 79.7% coverage.
  • Live: patched a running simplyblock-monitoring Deployment on cluster91 with the fixed command; two-node-event-collector container came up Running and logged Two-node event collector started cleanly against real FoundationDB/RPC.
  • Marked draft pending sbcli's translate-go (#1471) landing, since this only matters once that image is what clusters run.

…eleted Go binary

monitoringServices() still named /usr/local/bin/sb-event-collector as the
two-node-event-collector container's command. sbcli's translate-go rewrote
it as simplyblock_core/services/two_node_event_collector.py and removed the
Go toolchain stage that built that binary, so every pod running this image
crash-loops on "stat /usr/local/bin/sb-event-collector: no such file or
directory" (confirmed live on cluster91). Point the service entry at the
module like every other one, drop the now-dead twoNodeEventCollectorBinary
const, and update the two tests that pinned the old binary path.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant