Capped processes can be attached late, released, and have their out-of-memory kills counted; still off.
Late attach, release and OOM accounting for memory-capped processes exist under the same off-by-default gating.
What's wrong with this entry?
Processes started before the memory cap is decided are parked and attached later along with their children, a process and its descendants can be released back out of the cap, and OOM kills are counted. Same gating as the rest of the tool cgroup work, which is off by default.
- The OOM counter is polled from
memory.eventsand reported asoom_kills, along withcgroup_v2, ontengu_tool_cgroup. - A separate keeper directory holds processes meant to outlive the capped tree.
released_count
Strings lifted out of the shipped bundle, so the claim above can be checked against them.
Related
Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.
-
v2.1.236
macOS sandbox re-applies read and delete denials inside writable folders
Both mention sandbox
-
v2.1.236
Sandbox proxy no longer writes to sockets that have gone away
Both mention sandbox
-
v2.1.236
IPv6 hosts through the proxy
Both mention sandbox