watchdogd: Add -k kenv option to store the process ID.
Needs ReviewPublic
Actions

Authored by ray on Mon, May 25, 8:02 AM.

Details

Reviewers

glebius
imp
pjd
jmg

Summary

Storing the PID in the kernel environment provides access to it when operating
on read-only filesystems. Additionally, it prevents issues with stale PID
files left behind if watchdogd is terminated by SIGKILL.

Diff Detail

Repository

rG FreeBSD src repository

Lint

Lint Skipped

Unit

Tests Skipped

Build Status

Buildable 73416
Build 70299: arc lint + arc unit

Event Timeline

ray created this revision.Mon, May 25, 8:02 AM

Herald added a subscriber: ziaee. · View Herald TranscriptMon, May 25, 8:02 AM

ray requested review of this revision.Mon, May 25, 8:02 AM

Harbormaster completed remote builds in B73416: Diff 178541.Mon, May 25, 8:02 AM

glebius added reviewers: pjd, jmg.Mon, May 25, 6:14 PM

To me this looks like a more generic problem rather than just the watchdogd's problem. Should every daemon have this feature? Or should we say that /var/run shall be mdmfs(8) on hosts with read-only root?

I added authors of the pidfile(3) to add people who definitely thought a lot on this problem.

In D57231#1311555, @glebius wrote:

To me this looks like a more generic problem rather than just the watchdogd's problem. Should every daemon have this feature? Or should we say that /var/run shall be mdmfs(8) on hosts with read-only root?

I added authors of the pidfile(3) to add people who definitely thought a lot on this problem.

IMO, it's good workaround for life critical services in the embedded area. But for generic use, instead of pidfile, we have to design proper subsystem.
Embedded devices have no screen to show an issue, so recover through wd's reset may help with many issues.