On Fri, Mar 10, 2006 at 02:30:27PM +0100, Lars Hartung wrote:
> 09:08:39 SetVSStatus: 0x2000001, client 162058, server 162058
> 09:08:39 SetVSStatus: 0x2000001, NewVS 162058, CBstatus 1
> 09:08:39 AllocRecord: No space left in volume log.
> 09:08:39 GatherFids: Entering for 0x2000001.267f.6141
> 09:08:39 GetResObjs: returning(0)
> 09:08:39 AllocRecord: No space left in volume log.
> ....
> 09:08:39 SpoolVMLogRecord - no space left in volume
> 09:08:39 - returns ENOSPC
> 09:08:39 ViceMakeDir: Error 28 during SpoolVMLogRecord for child

Do you have a lot of directories?

Each directory will use one resolution log entry as an anchor point
(common ancestor) for server-server resolution. But when volumes are
created they only get about 4096 resolution log entries. When we run out
the servers can't keep track of new updates.

You can bump this limit up with volutil,

    volutil setlogparms 2000001 reson 4 logsize 32768

This would set the logsize to 32768 (32K entries) for volume replica
2000001. You'd need to do this for each volume replica in a replicated
group.

> Should i raise the size of the logfile? 
> During installation it is said that 20MB should be sufficient.

Yeah, 20MB definitely should be sufficient. I've not seen a server that
updates close to that amount of file/directory metadata in a single RVM
transaction.

Jan