hive-agent-sock: kill what a subagent run leaves in its cgroup after it ends
This commit is contained in:
parent
566758483d
commit
2689efebcf
1 changed files with 82 additions and 11 deletions
|
|
@ -44,6 +44,8 @@ a crate of its own.
|
|||
`hive-bash-mcp` puts each bash task in `<unit>/tasks/<id>/` and
|
||||
`hive-subagent-mcp` each subagent run in `<unit>/runs/<name>-<n>/`, both with
|
||||
`memory.oom.group=1`, so the OOM killer takes one whole run and leaves the
|
||||
daemon alone. Both daemons are delegated their cgroup with `Delegate=memory
|
||||
daemon alone. A subagent run's cgroup is also killed (`cgroup.kill`) if
|
||||
processes are still in it 30 seconds after the run ended; a bash task's is
|
||||
left alone. Both daemons are delegated their cgroup with `Delegate=memory
|
||||
pids` and `DelegateSubgroup=daemon` (`nix/agent-modules/mcp.nix`); outside such
|
||||
a unit the module logs one warning and runs stay in the daemon's cgroup.
|
||||
|
|
|
|||
Loading…
Reference in a new issue