EVA5000 to EVA8000 data replication problems

[Andrej Jerina]

We start replication from EVA5000 to EVA8000, after normalization and few days of normal operation TRU64 cluster practicaly hanged (only ping worked). In ES45 console wwidmgr also need very long time to return results and disk were not seen.
After 2 hours EVA5000 rebooted each controller and everything become normal. After few days at same event we simply powered off EVA8000 and after few seconds EVA5000 normalized. Vdisk that were not in DR Groups responds normaly all the time.
SAN is simple topology and both EVAs are on same switch. EVA5000 is crossed and EVA8000 is straight connected.
We also set Aptpolicy to 1, no effect.
fabric 1 with 2 4/32 FC SW (fw 5.0.3)+ 1 1/16 FC SW( fw 2.6.2)
fabric 2 with 2 4/32 FC SW (fw 5.0.3)+ 1 1/16 FC SW( fw 2.6.2)
EVA5000 3.028
EVA8000 5.110
CV 5.0.0.208
near time of event we can see in event log " Unknown Indicated Virtual Disk has transitioned to Stalled Too Long " and sometimes also "The Data Replication Path between this Site and the Alternate Site has closed, possibly due to a connection failure between the specified host port and the Alternate Site"
Any ideas?
Regards
Andrej

[Andrej Jerina]

Some more info
It looks that this is only TRU64 Cluster and CA related, other nodes with replicated disks works normal.
If cluster nodes are booted and halted before came up completely, boot disks are not accessible in console until replication is disabled or after appr. 2 hours EVA5000 automaticaly reboots its controllers.

[Andrej Jerina]

This was CA problem and only TRu64 had real problem with, all other hosts survived.
For CA I zoned together 1st 4 ports of EVA8k and 4 ports of EVA5k, other 4 ports of EVA8k are used for host conection. After that change there are no more problems or errors on EVAs.