FC Tape drives unaccessible

[Reiniche]

Hi there,
We’ve problem accessing tape drives connected to one of the fabrics we’re using. Actually, none of the tape drives connected to first fabric are viewable from our servers, whereas at the same time, the same servers, using the same type of connection (HBA, fibre, SFP, switch, topology…) can access to the other half of tape drives, through the second fabric.
Our SAN is based on 4x switches (StorageWorks 8/80) split into 2x physically independent fabrics (no ISL link between the 2 fabrics). Both fabrics are (almost) symmetric (connections, zoning, FW level…) and the targets accessible through one fabric, are also accessible through the other one. The only known difference is that half of our drives are accessible only through one fabric (HP Ultrium LTO4) whereas the other ones are connected to the second fabric.
We’ve swapped HBA, links, SFP, etc… We’ve even tried to force the speed on the HBAs and the switches, without any kind of improvement… Do you have any idea?
Any help would be most welcome...
Thx -

[The Brit]

What OS are your servers running??

Have these tapedrives ever been seen from the server?

or are they replacements for malfunctioning devices?

[Torsten.]

There must be a library housing these drives. Depending on the library model (e.g. ESLe, EML, ...) did you present the drives to servers using command view TL?

[Reiniche]

We’ve finally found out what is going wrong with one of our fabrics:
Both fabrics are using dual ISL links in failover mode (no trunking) It appears that one of those links is defective (either the SFP or the fibre, we don’t know yet) The tricky point is that the link was good enough to handle disks access, but not for tape drives (streaming devices…)
The only way to discover this failure was to disable the link, and not to swap it with the other ISL links, as we did in the past. Since it is disabled, we managed to rediscover properly all the tape targets on all hosts.
Thx -