* Managing difficult log files
@ 2019-05-17 1:03 Dewayne Geraghty
2019-05-17 9:08 ` Laurent Bercot
0 siblings, 1 reply; 2+ messages in thread
From: Dewayne Geraghty @ 2019-05-17 1:03 UTC (permalink / raw)
To: supervision
I would appreciate advise as to whether these are acceptable questions
for this list?
I would like to use the s6-log features for applications that have
difficult log management. In this instance, apache. From
procstat –f `pgrep httpd`
we're advised that each http has write-append to the ErrorLog, and if we
use s6-log then we have locking problems.
So an intermediate reader appears to be a sound workaround. Does anyone
have a better solution than
ErrorLog “|'execlineb –Pc fdmove 1 0 s6-fdholder-retrieve
/s/live/servicedirs/s6rc-fdholder/s ‘pipe:s6rc-w-final-logger’ fdswap 0
1 sh -c "read X; echo ${X}"
Please note the above is for testing, as this will be moved into a
script where logs are scanned for anomalies.
Previously I tried:
ErrorLog "|/usr/local/bin/s6-log -b n32 s10000 S7000000 T !'/usr/bin/xz
-7q' /var/log/httpd-error"
I'm looking at using a similar approach to samba and others as I dig
into this.
Kind regards, Dewayne
PS Please be tolerant as manipulating descriptors is very new to me, but
a lot of people struggle with managing apache logs and this would be a
nice-to-have use of s6-log. (Even if the users demand daily logs via
cron & 59 23 * * * s6-svscan -a apache24-log ;) ). Now to samba...
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: Managing difficult log files
2019-05-17 1:03 Managing difficult log files Dewayne Geraghty
@ 2019-05-17 9:08 ` Laurent Bercot
0 siblings, 0 replies; 2+ messages in thread
From: Laurent Bercot @ 2019-05-17 9:08 UTC (permalink / raw)
To: supervision
>So an intermediate reader appears to be a sound workaround. Does anyone
>have a better solution than
>
>ErrorLog “|'execlineb –Pc fdmove 1 0 s6-fdholder-retrieve
>/s/live/servicedirs/s6rc-fdholder/s ‘pipe:s6rc-w-final-logger’ fdswap 0
>1 sh -c "read X; echo ${X}"
1. "cat" should work instead of your shell excerpt, shouldn't it?
Isn't it supposed to flush on every line?
2. Instead of spawning a separate process for each ErrorLog, it
should be possible to create a fifo, and declare that fifo as your
ErrorLog file. Every httpd would then write into the fifo, and you
would have a single s6-log reading from it.
--
Laurent
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2019-05-17 9:08 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-05-17 1:03 Managing difficult log files Dewayne Geraghty
2019-05-17 9:08 ` Laurent Bercot
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).