Coda File System

Re: Replication server problems

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Mon, 18 Dec 2006 12:36:32 -0500
On Mon, Dec 18, 2006 at 10:35:05AM +0100, Achim Stumpf wrote:
> On clusty2 (slave):
> # cat vicetab
> clusty1.mytest.de   /vicepa   ftree   width=256,depth=3
> clusty2.mytest.de   /vicepa   ftree   width=256,depth=3

That looks good.

> The SrvLog of clusty2 (slave):
> Date: Mon 12/18/2006
...
> Partition /vicepa: inodes in use: 0, total: 16777216.
> 10:15:52 Partition /vicepa: 8211208K available (minfree=5%), 8191888K free.

And the server on clusty2 recognized the vicepa partition. So something
else must be going wrong here.

> After that I have restarted the servers on both machines.
> And on clusty1 (SCM):
> # createvol_rep coda.rep clusty1.mytest.de/vicepa clusty2.mytest.de/vicepa
> egrep: /tmp/vollist.1947: No such file or directory
> egrep: /tmp/vollist.1947: No such file or directory
> grep: /tmp/vollist.1947: No such file or directory

Those failing greps are a good indication that the following command
failed,

    volutil -h "$SERVER" getvolumelist "/tmp/vollist.$$" 2>/dev/null

I guess we won't see the errors since stderr is redirected to /dev/null.
So I wonder if there is an error message when we run the following on
clusty1.

    volutil -h clusty2.mytest.de getvolumelist

This should dump the list of volumes (and partitions) to stdout and if
it doesn't it should give some error message. My guess is that there may
a slight difference for /vice/db/volutil.tk on both machines and so the
volutil on clusty1 fails to authenticate with the server on clusty2.

Jan
Received on 2006-12-18 12:38:56