s6-permafailon not acting as expected

From: Xavier Stonestreet <xstonestreet_at_gmail.com>
Date: Mon, 16 Nov 2020 19:01:18 +0100

Hello,

I'm seeing unexpected results with s6-permafailon.
I have this line in my finish script:

s6-permafailon 120 2 1-255,SIGBUS,SIGSEGV exit 0

If I kill the service twice in a row with s6-svc -t, s6-permafailon
triggers a permanent failure with the message:

s6-permafailon: info: PERMANENT FAILURE triggered after 2 events
involving exit code 0 in the last 120 seconds.

The service did die with exit code 0, but s6-permafail is wrong, since
exit code 0 is not specified in the filter.

s6-svdt reports:
2020-11-16 16:00:17.451308493 exitcode 0
2020-11-16 16:00:28.495276767 exitcode 0

Looking at the source code in s6-permafailon.c, it seems to me that
the codes[32] bitarray is not initialized with zeroes before calling
bitarray_set(n). But not being intimately familiar with the code I'm
probably looking at it incorrectly. And I'm not instrumented nor
knowledgeable enough to run a debugger.

Could you shed some light on this unexpected behavior and tell me what
I'm doing wrong, as the case may be.

Great supervision and service management toolbox by the way - I love it! :)

Thanks,
X.

s6 2.9.2.0
skalibs 2.9.3.0
execline 2.6.1.1
built from source, linked with musl libc on Linux
Received on Mon Nov 16 2020 - 18:01:18 UTC