Mike Christie <michaelc@...> writes:

> 
> If you see the above sendpage error on the target and only see a conn
> error 1020 on the initiator side, then it might be the target got a
> failure during a send operation and dropped the connection. 1020 on the
> initiator side indicates the target dropped the tcp/ip connection.
> 

Yes, that is what it looks like. The tcp send operation on the target
seems to fail.

> 
> Ok. It seems the connection was dropped and the initiator was able to
> relogin and continue IO.
> 

That's good to know. I guess the TCP mechanism re-transmit the failed
packets so this should be nothing to worry about.

> 
> If the problem is the target dropping the connection because it could
> not send IO, then I am not sure.
> 
> You might be sending more IO than the target can handle and in that case
> you can lower the targets per session queue depth if LIO has a setting
> like that. If not then on the initiator side you can lower it by setting
> node.session.cmds_max lower. To set that either set it in iscsid.conf
> then rediscover the target and relogin or run iscsiadm -m node -o update
> to update it for specific targets already setup then relogin (probably
> also want to set it in iscsid.conf for future targets/portals too if it
> works).
> 

Currently the cmds_max is set to 128 by default, I set it to 64 but the
send page errors on the target still happen with the same frequency. I
couldn't find any setting on the LIO target configuring the queue depth.
I will look into it some more and see if there is any setting like that.

Thanks!
Amit

-- 
You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/open-iscsi?hl=en.

Reply via email to