[OpenAFS] Re: Ubik problem

Atro Tossavainen atro.tossavainen+openafs@helsinki.fi
Tue, 13 Apr 2010 08:43:47 +0300 (EEST)


> We had this problem with our DB servers here. It would appear that you
> also need to specify -rxbind, to stop the servers from sending packets
> from interfaces that are in NetRestrict.

Which command(s) should be started with this flag?

(Remember is still on IBM AFS and probably does not
 support -rxbind if it's something that OpenAFS implemented post IBM.)


> Atro's problem appears to be that they are advertising the extra
> addresses, though, not that they're sending packets out over them.

What I don't get is why this would have changed transparently on the
sun4x_58 server when I changed the *other* db server from sun4x_58 to
sunx86_510 and from IBM AFS to OpenAFS.


> The OpenAFS NetRestrict documentation does not mention the use of a
> preceding 'M'.

Neither does the IBM documentation, which is what I said, I think.

> I suspect the change IBM implemented was never passed on to OpenAFS
> and as far as I can tell from the source its presence will invalidate
> the address and prevent it from being used as a filter.

On the IBM AFS sun4x_58 server where it was explicitly recommended
by IBM AFS support in early 2005? :-)  I'm not running the OpenAFS
db servers.  In fact, I seem to be running a mishmash of IBM AFS
servers - the various components aren't even all the same version.

>From my reading of IBM AFS patch READMEs, it appears that "M" appeared
in AFS 3.6 Patch 13 (aka Build Level 2.57) (APAR IY77101).  This seems
to have been released in December 2005, but IBM support have indicated
its use in a support ticket conversation in January 2005 already.  It
did not work, and in connection with another support ticket in September
2005, IBM made available a patched version of the 3.6 2.56 *fileserver*
only where it finally did work.  So I don't think I've ever had any
database servers that are even supposed to obey "M" in NetRestrict,
but at the same time, I haven't had a problem with this so I haven't
noticed.  What the...

Atro Tossavainen (Mr.)               / The Institute of Biotechnology at
Systems Analyst, Techno-Amish &     / the University of Helsinki, Finland,
+358-9-19158939  UNIX Dinosaur     / employs me, but my opinions are my own.
< URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS