[2005-08-08 15:09:57] WARN connection/_accept_connection accept()
failed with error 24: Too many open files
I'm getting this error I had earlier once again. I reported this on
June 1, and restarted the server at that time. it's been working fine
again since then, but now the error has shown up again. lsof lists a
lot of lines like this:
icecast 22212 nobody 141u sock 0,0 1363923053
can't identify protocol
icecast 22212 nobody 142u sock 0,0 1363923057
can't identify protocol
icecast 22212 nobody 143u sock 0,0 1363923061
can't identify protocol
They are present for all the pid's used by icecast. There are 22
icecast pids on the system. I checked in the /proc/pid/fd directory for
each pid, and each one is filled with exactly 1024 file descriptors.
I don't want to leave the server in this state for too long, but if
there is any information I can provide within the next hour or so from
this current state I'm happy to do some testing. but it's no longer
accepting connections, so I can't leave it down much longer.
As I said last time, I'm running Slackware Linux 10.0, kernel 2.4.26 and
icecast 2.2. We usually have less than 100 clients connected, but I
have as many as 13 sources connected at once. Only 6 are heavily used,
and 6 more are just fallbacks only used when the primary 6 go down.
There was no suspicious behavior I could find in the logs. All looked
just as it normally does. I suspect this happened over time since this
instance of icecast has been running non-stop since June 1.
Thanks,
Joel Ebel
