[OpenAFS] accessing volumes with missing replicas hangs eternally.

[Matthew Andrews]

Hello, I noticed a slightly odd behavior on an openafs client.

In the process of upgrading fileservers in my test environment, I
accidentally forgot to delete RO volumes from the fileservers I was
upgrading and recreate them on the new fileservers.

in this situation(no RO replicas are available) an afs3.4a client on
sol2.5.1 will time out and give an error about loseing contact with the
missing fileserver. this is the behavior I expect.

on a redhat 7.0 machine running openAFS 1.0.3, however trying to access
files on the missing volume simply hang forever.

for the afs3.4a case, the client thus became happy when I simply removed
the old replicas and created new ones on the new fileserver. the openafs
client required a reboot befor it could access the affected volumes.


is this the expected behavior? if so, why? The ibm behavior seems to
provide for more gracefull recovery of the client.

-Matthew Andrews