Coda File System

Re: SCM cache overflow

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Fri, 21 Nov 2003 10:26:25 -0500
On Fri, Nov 21, 2003 at 09:38:58AM +0100, Fr?d?ric LASSALLE wrote:
> Coda-Master:~# cfs whereis /coda/coda-master.intranet.lan/
> Coda-Master.intranet.lan
> 
> Coda-Slave:~# cfs whereis /coda/coda-master.intranet.lan/
> Coda-Master.intranet.lan

The root volume is not replicated.

At the moment is it very messy and error prone to extend the replication
group of an already existing volume. The easiest thing to do is to
create a new volume that is replicated.

Check whether you have a 'VSG' number for both servers in the VSGDB, if
not add one,

    /vice/db/VSGDB
    E0000101 coda-master.intranet.lan coda-slave.intranet.lan

Then create a replicated volume that we will use as the replicated root.

    createvol_rep codaroot.rep E0000101

Then replace the name of the rootvolume in the 'ROOTVOLUME' file,

    /vice/db/ROOTVOLUME
    codaroot.rep

Finally, the client has the wrong things cached, so you will have to
restart and reinitialize it.

    # killall venus
    # umount /coda
    # venus -init &

Now cfs lv /coda/coda-master.intranet.lan should show that the
rootvolume is named 'codaroot.rep'. And cfs whereis will return both
coda-master.intranet.lan and coda-slave.intranet.lan.

Jan
Received on 2003-11-21 10:30:42