Does the iSCSI initiator fully clean up failed connections?

The situation I have (as before) is a S10U3 initiator talking to a
snv_54 target. Thanks to Rick I got this working.

But I've been having random network problems since, and
realized today what the cause was.

I'm seeing the client use a large number of tcp ports. From
netstat -an on the machine with the initiator (10.1.1.110),

10.1.1.110.35760      10.1.1.124.3260      64240      0 64240      0
CLOSE_WAIT
10.1.1.110.33258     10.1.1.124.3260      64240      0 64240      0
CLOSE_WAIT
10.1.1.110.60834     10.1.1.124.3260      64240      0 64240      0
CLOSE_WAIT
10.1.1.110.58261     10.1.1.124.3260      64240      0 64240      0
CLOSE_WAIT
...

there are 32703 of these, plus one ESTABLISHED. On the target end
(that's 10.1.1.124), I just see the single ESTABLISHED connection.

My guess is that these might be a hangover from my earlier failed attempts
to communicate (when iscitgtd was dumping core every time) but even then
shouldn't it clean these connections up?

In the end I had to resort to a reboot, as I hadn't got enough free
ports for any other applications to work.

Since the reboot it's been fine, so it doesn't seem as though it's leaving
all these connections in CLOSE_WAIT in normal operation.

--
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to