mds9509 rscn rejected

gentilucci · ‎06-30-2008

Hi

I have the following messages:

2008 Jun 28 11:28:06 RSCN 2 RSCN_REJECT 15 RSCN rejected: Received LS-RJT: fcid 0x6d0038, Reason code-Unable to perform command request, Reason code expl-Insufficient resources

2008 Jun 28 11:28:06 RSCN 2 RSCN_REJECT 15 RSCN rejected: Received LS-RJT: fcid 0x6d003e, Reason code-Unable to perform command request, Reason code expl-Insufficient resources

Mainly on AIx hba (IBM host)on different fc-id-.

The new OS now is 3.2.3a.

We lost also one path (of diffrent system) to a disk (HDS USP).

Any idea?

Many thanks in advance.

Walter

tblancha · ‎07-02-2008

Yes, it mean that those two devices rejected the RSCN. So, something happened in the SAN that these two devices are zoned with. So, a registered state change notification was sent to these devices. They rejected it and in the rejection, they indicated why they were rejecting was because of insufficient resources. The MDS will try again and eventually the HBAs will accept.

gentilucci · ‎07-03-2008

Hi

thanks for answer.

Is there a way to understand exactly wich is the source of this problem?

Also, I can't understand what "insufficent resource" means.

Maybe wrong setting on hba? Or a setting on MDS?

The problem is that some hosts lost the path.

If I look into show tech I can see that the RSCN comes only from Emulex (IBM) hba; they have lot of IBM host with AIX.

The configuration is:

2 vsan with two pair of 9509 in ISL.

In one Vsan all is running good on the other we got the problems.

Bye

Walter

stephen2615 · ‎07-06-2008

Walter,

Did anyone do something to the switch when this happened? In particular, did anyone shove in a new blade at around the time these events occurred?

Did someone reboot one of the switches?

Stephen

tblancha · ‎07-07-2008

This is probably not the problem and rather just a distraction. The HBA is reporting that it has insufficient resources at that time but the RSCN gets retransmitted and so it will eventually flow across to the HBA. The key is that there are RSCN's flowing. RSCN flowing in of itself is not a bad thing- it is just a frame indicating that something changed. If you are reporting that outages occured during that time, then it's probably a bad thing. So, there is a feature known as the fcanalyzer that is onboard the MDS supervisor that will give you some clue as to what is going on. Depending on how much traffic there is and what vsan it is in, you will want to do a display filter. Also, we have the ability to span ports. There are also many show commands which will show what's going on. You might want to open a TAC case and they can give you the specific commands that are best for your situation.

marsnow · ‎07-27-2008

This is a known problem with AIX. You can turn off RSCN to an interface if they are causing the host problems. Hidden command.

rscn suppress interface fcx/x

Beware before turning RSCN off as this may cause your host other problems as it will not know when device states change in the SAN. You should look at fixing the host and turn off RSCN as a last resort.