1888965 - Spurious "WARNING: process [pid] is a zombie: file [...]/process_util_posix.cc" log messages

Status: RESOLVED DUPLICATE of bug 1752638

(Core :: IPC, defect)

Description

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

Noticed after bug 1881386 landed: the log message complaining about finding a child process in the zombie state is happening during normal operation in a non-container environment. I've seen this locally, and the processes in question do seem to have eventually ceased to exist.

This is a little weird, because in this case the fork server is still running and has SIGCHLD set to SIG_IGN (bug 1875409), so its child processes should be auto-waited immediately after they exit. Maybe there's a race window here where the process can be observed as a zombie in the middle of that procedure, and we're hitting it sometimes by chance.

This seems like it's just a cosmetic problem — the process is definitely going away, so it's okay to stop waiting for it, which is what the zombie check is for — but there should be a bug on file for it.

This is another issue that bug 1752638 should address.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1881386

Comment 2

[Donal Meehan [:dmeehan]]

:jld next week is the final week of nightly before Fx126 hits beta.
based on Comment 0, will this be a wontfix for Fx126?

Comment 3

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

Yes, wontfix, and… on second thought, this doesn't really need to be classified as a defect, I don't think.

Comment 5

[Daniel Holbert [:dholbert]]

(In reply to Jed Davis [:jld] (on leave until 08-31) from comment #3)

Yes, wontfix, and… on second thought, this doesn't really need to be classified as a defect, I don't think.

Just clarifying -- this was wontfix for 126, but not wontfix indefinitely, right?

RE this not being a defect -- I disagree... previously, you could start Firefox and quit without anything at all logged to stdout/stderr, under normal conditions. Now, you're ~guaranteed (on my machine at last) to get one or several copies of this warning (per bug 1912432). I think we try pretty hard to keep opt builds ~silent by default in terms of terminal output, so from that perspective, this issue is a regression and a defect. (Previously you had to flip a pref to see this issue, but that pref is now on-by-default for Nightly, as of bug 1874689).

At a minimum, it seems like we should change this logging to be debug-build-specific, or something along those lines (assuming it's non-concerning)...

--> Tentatively reclassifying as 'defect' per above, since I think we should consider it as-such (albeit a minor one) now that all Nightly users are affected by-default. (Sorry if this is presumptive; happy to have this assessment reversed if appropriate, but I still think we should take this step away from ~silent terminalspew seriously.)

Comment 7

[Nika Layzell [:nika] (ni? for response)]

As previously noted, this should be fixed by bug 1752638, so closing this as a dupe for now.