Re: 14 Jan 2021 08:25:36 GMT Breakage in RX Connection ID calculation

P.S.: I get timeouts with ssh logins even between patched 1.8.6 clients during aklog (tested by commenting aklog from pam settings -> no timeout). Token was obtained after timeout.

Cheers and thanks again,

Ado

Heinz-Ado Arnolds wrote on 14.01.21 18:45:
> Dear Jeffrey,
> 
> many thanks for your fast response from Germany too!
> 
> When issuing "vos listvol <server>" on a patched 1.8.6 client to an 1.6.22.1 <server>, I still get "Could not get the list of partitions from the server. Possible communication failure". The same command works from a client running 1.6.23.
> 
> I'm still having problems when doing an ssh from a patched 1.8.6 client to a server running an unpatched 1.8.6 and vice versa. The login process hangs during aklog. That means both machines have to run a patched 1.8.6?
> 
> Thanks and cheers,
> 
> Ado
> 
> Neil Brown wrote on 14.01.21 18:19:
>> On Thu, 14 Jan 2021, Jeffrey E Altman wrote:
>>
>>> Patches to correct the flaw are available from OpenAFS Gerrit
>>>
>>>  https://gerrit.openafs.org/14491
>>>  rx: rx_InitHost do not overwrite RAND_bytes rx_nextCid
>>>
>>>  https://gerrit.openafs.org/14492
>>>  rx: update_nextCid overflow handling is broken
>>
>> Jeffrey,
>>
>> Informatics at the University of Edinburgh, are very grateful for you and AuriStor's quick response to this, as I'm sure are many other members of this list.
>>
>> Also thanks, to the other members of this list with their efforts in diagnosing the root cause.
>>
>> Hopefully we'll be able to try a patched version tomorrow.
>>
>> Thanks again,
>>
>> Neil
>