[OpenAFS-devel] openafs-1.3.76: Multihoming seems broken

[Niklas Edmundsson]

On Sat, 18 Dec 2004, Harald Barth wrote:

>> However, openafs 1.3.76 (which is the one I'm pounding on) seems to
>> flag the whole server as down when it can't reach an interface:
>>
>> # ./fs checkservers
>> These servers unavailable due to network or server problems:
>> chosan.hpc2n.umu.se.
>>
>> Console output:
>> afs: Lost contact with file server 130.239.45.228 in cell hpc2n.umu.se
>
> Server is 1.2.13 (says rxdebug). Does it log anything about your client?

Should be correct.

> Your client should send packets to your server to poll if it is still
> down. Can you see these?

Yup, this is what tcpdump captures during a "fs checkservers" run:

14:53:33.071839744 n11.afs3-callback > chosan.afs3-fileserver: udp 32
14:53:33.072096000 n11.afs3-callback > chosan-k.afs3-fileserver: udp 32
14:53:33.072367872 chosan.afs3-fileserver > n11.afs3-callback: udp 36 (DF)
14:53:33.073500416 n11.afs3-callback > chosan.afs3-fileserver: udp 65
14:53:35.158586368 n11.afs3-callback > chosan-k.afs3-fileserver: udp 32
14:53:37.358892800 n11.afs3-callback > chosan-k.afs3-fileserver: udp 65
14:53:37.558925568 n11.afs3-callback > chosan-k.afs3-fileserver: udp 32
14:53:39.559187712 n11.afs3-callback > chosan-k.afs3-fileserver: udp 65
14:53:40.360304896 n11.afs3-callback > chosan-k.afs3-fileserver: udp 32

>> I'm not on the list.
>
> UUUh, you can fix that ;-)

Yeah, I had hoped not having to subscribe just to report a problem...

/Nikke
-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
  Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se     |    nikke@hpc2n.umu.se
---------------------------------------------------------------------------
  Operator halted! Startrek's on!
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=