Russ Allbery via RT wrote:

>> Just to be clear, the problem happens when the ticket cache is
>> refreshed. I.e., the tickets for the existing SASL/GSSAPI connection
>> hadn't actually yet expired, just the ticket cache was refreshed with
>> new tickets.  I can understand why the SASL/GSSAPI context would be
>> closed out on *expiration* but I think a refresh shouldn't have this
>> effect. ;)

If it is possible, can you post a stack trace at the point the context
is deemed to be invalid?

That would help a lot.