I've been doing a little research on this problem, and I see some scattered reports of similar problems over the past 3-4 years.<br><br>I have asterisk running on a cloud machine, so far fairly low volume, and every couple weeks, it gets <br>
clogged up with open files until it can't open new files any more. It's not dead, and the CLI is working fine.<br>It's just reached the limit.<br><br>
I'm running 1.6.2 on a fairly late-model kernel (2.6.24).<br><br>
An 'lsof -p <pid>' on asterisk at that point yields around 700 lines of entries that look like this:<br><br>asterisk 25361 root 27u IPv4 1701345 UDP *:16932 <br>asterisk 25361 root 28u IPv4 1701346 UDP *:16933 <br>
asterisk 25361 root 29u IPv4 1701926 UDP *:18054 <br>asterisk 25361 root 30u IPv4 1701927 UDP *:18055 <br>asterisk 25361 root 31u IPv4 1701965 UDP *:13798 <br>
asterisk 25361 root 32u IPv4 1701966 UDP *:13799 <br><br>The list must grow slowly with time, as when I checked today, the number of open files<br>was at around 700 or so, and so I had another 300 or so descriptors to go. I restarted<br>
the server to prevent the total clog up, which is very inconvenient for users.<br><br>Now, I could be way, way off, but the UDP and port numbers between 10K and 20K<br>kind of make it look like RTP connections are not getting properly closed. Could this<br>
be the reason?<br><br>Anyone else seeing this? I'm only having problems on one machine (so far). <br><br>murf<br><br clear="all"><br>-- <br>Steve Murphy<br>ParseTree Corp<br><br>