https://bugs.koozali.org/show_bug.cgi?id=12298
Jean-Philippe Pialasse <[email protected]> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |IN_PROGRESS
Resolution|FIXED |---
--- Comment #17 from Jean-Philippe Pialasse <[email protected]> ---
(In reply to Gary Douglas from comment #16)
> don't know if this is intentional but
> etc/e-smith/templates/etc/opt/remi/php81/php.d/20-opcache.ini/10base has;
> opcache.file_cache= /var/opt/remi/php74/lib/php/opcache
good catch
but should not create any issue, except not intended to be this way
this path will only be user by NC when called ny command line
>
> Had some fun with opcache.so segfaults again after php updates this morning,
> so custom-templated this to /php81/ upped buffer + memory + files, set
I experimented today 502 bad gateway issue with nc26 and php81 and what i
suspect to be opcache issue, but not able to reproduce it easily, sometime it
keeps failing to the point httpd errors while trying to proxy requests. It only
start for me after today update of php81 (php81-php-opcache.x86_64
8.1.23-1.el7.remi). no similar issue in the last weeks (neither with php81 nor
php74)
[WARNING] [pool php81-nextcloud] child 12122 exited on signal 11 (SIGSEGV)
after 310.021004 seconds from start
issued a couple restar of httpd and php81 without change and a ultimate restart
did fixed it and no more since.
I have also a lot of this in messages log
Aug 31 05:23:28 atlas kernel: [6382273.317088] traps: php-fpm[20041] general
protection ip:7fecfc1ee3bc sp:7ffe163654d0 error:0 in
opcache.so[7fecfc1e7000+b8000]
Aug 31 05:23:28 atlas kernel: traps: php-fpm[20041] general protection
ip:7fecfc1ee3bc sp:7ffe163654d0 error:0 in opcache.so[7fecfc1e7000+b8000]
one issue i see is in opcache setting file template we have
opcache.interned_strings_buffer= {$nextcloud{'opcache.interned_strings_buffer'}
|| 32;}
opcache.max_accelerated_files= 10000
opcache.memory_consumption= {$nextcloud{'opcache.memory_consumption'} || 128;}
and in nextcloud pool we have
my $opcacheBuff = $nextcloud{'opcache.interned_strings_buffer'} || 16;
my $opcacheMem = $nextcloud{'opcache.memory_consumption'} || 128;
php_value[opcache.interned_strings_buffer] = 16
php_value[opcache.max_accelerated_files] = 10000
php_value[opcache.memory_consumption] = 128
so default are lower in the pool... which might lead to the segfault
but was not able to reproduce in reliable way
however when the issue arise, I can make it stop by simply
rm -rf /var/lib/php/nextcloud/opcache/*
systemctl restart php81-php-fpm.service
so seems it is something like corrupted cache...
we might need to add a routine to clear this path regularly
> additional NC config.php Redis settings + memcache.distributed + dbindex =>
> 0 + timeouts => 1.5 per Nextcloud caching recommendations.
redis is another issue see https://bugs.koozali.org/show_bug.cgi?id=12401
I do not think it is related is anyway with the proxy error and php pool sig 11
--
You are receiving this mail because:
You are the QA Contact for the bug._______________________________________________
Mail for each SME Contribs bug report
To unsubscribe, e-mail [email protected]
Searchable archive at https://lists.contribs.org/mailman/public/contribteam/
