Alan said:
> Instead of querying the control unit, you need to QUERY PAV ALL.
HyperPAVs
> are not associated with a base device until an I/O is performed, and
that
> association only lasts for the duration of that I/O. Hence the failure
of
> QUERY CU DASD ALIAS to give you anything useful. Inste
Samir,
Stefan mentioned:
> When Linux is running in an LPAR, then it enables PAV itself,
> and everything should work as you expected. In case of a z/VM
> system however, it is z/VM that enables PAV, and in your case
> it enables Hyper PAV. Perhaps it is possible to configure the
> use of base PA
Hi Ray,
I can't speak to the operation of sync'ing the two EKMs, but one thing
comes to mind about why you can't encrypt the tapes: Does the CU know the
IP address(es) of the EKM(s) to be used, so that it can perform the
required handshakes during encryption operations? The sense data you
suppli
It's worth noting that the underpinnings of SCSIDISC changed significantly
in 5.4.0. In previous z/VM releases, a file named RXSCSIFN needed to be
renamed to RXUSERFN to allow this tool to work. This is no longer the
case, and SCSIDISC expects the module to be named RXSCSIFN. As such, the
SCSIDI
Hi Susan,
> With CUIR, CP tells the control unit it may not take the last path
offline.
> What happens when CUIR is disabled? What happens if it is enabled, but
> z/VM does not support it? Does the CE have to intervene to prevent the
> last path from being taken offline?
If CUIR is disabled, th
Hi Susan...
> Whichever way the path is
> taken offline, if I lose access to the device because a path is taken
> offline (either with CUIR or manual), I have a problem. I think I'll go
> force a path off to my EDEVs and see if it recovers.
I agree with you that the multipathing of the EDEVs is
You're right, Pieter. The multipathing support in both Linux and z/VM
(EDEV) should allow each host to react to errors and outages that occur on
particular paths during CUIR-like events, but I think CUIR itself only
applies to CKD DASD. I'm not aware of anything in the SCSI standards that
would p
It is the SCSI driver stack's responsibility to perform the path
selection, which in most cases uses a straightforward round-robin
approach. I think the main difference worth highlighting is that of the
DS6000 path selection, which tries to respect the PREFerred/NOTPREFerred
settings in the EDEVic
