07-24-2016 11:45 AM
We have had this virtual fabric set up for years. For reasons as yet unknown the DCX-4S failed over unexpectedly. Everything came back ok on the default 128 fabric, but on our FICON virtual fabric the Mainframes themselves simply won't connect. Their ports show as In_Sync or No_Sync and the ports rack up link_fail & loss_sync counts.
We did an upgrade earlier in the year, (to v7.4.0a), which obviously involved CP reboots and failovers and saw no problems.
It is like some key FICON setting has been dropped, but I can't work out what!
Solved! Go to Solution.
08-10-2016 12:52 AM
Based on those information, it will be important to find the root cause of the unexpected failover and the unexpected consequence on several ports.
I have not seen similar issue. The recommendation will be to escalate this issue to the support team if you have necessary logs (supportsaves) that showing the issue.
Have you already open a case for that issue through your OEM if any?
08-10-2016 01:22 AM
Does failover occured during supportsave activities?
08-10-2016 02:59 AM
Unfortunately we do not have the ideal support for these. The supportsave did not reveal anything but did not cause any problems, (and to be honest the initial failure did not cause big issues).
I have since failed this over 3 times but have been unable to recover the FICON paths in this virtual fabric. All looks ok with FICON security but the Mainframe indicates that it doesn't like what it sees.
This is not critical as we are moving to a new solution on dedicated FICON switches soon. I suspect that if I destroy the virtual switch and then recreate it from scratch that it would resolve this. I am tied up with other stuff at the moment but I will try this just to satisfy my curiosity.
08-10-2016 05:26 AM
Does concerned ports are still enable?
Can you run a simple portenable on them if not done?
08-10-2016 09:55 AM
Disable does no good. Regardless of whether these are disabled or not the state swaps between In_Sync and No_Sync, but we see lots of SYNC LOSS messages, hence we keep them disabled.
When trying to vary the PCHID online from the processor itcomes online for a second and then MVS rejects the connection.
09-01-2016 12:24 AM
Have you already open a ticket to your OEM about this issue?
09-01-2016 12:39 AM
Sadly we don't have OEM support for this.
I will actually close this down now - I am pretty sure that this would be resolved by trashing the virtual switch and re-creating it, but we are due to be retiring this virtual fabric soon and moving to our dedicated mainframe switches.
This was set up for our tape access before we used FICON for disk. Since then we have implemented FICON disk on switches dedicatyed to this mainframe environment. We are now starting to replace our tape solution with one that willbe hosted on the dedicated mainframe switches, so this troublesome virtual fabric will die soon.