On Tue, May 27, 2003 at 06:09:07PM +0100, Malte Grosche wrote:
> Dear Jan,
> 
> Thank you very much for your help. It seems to have worked. First I have 
> checked the status of the log partition using rvmutil, status:
> 
> Status of log:           /dev/hda7
> 
>   status last written:   Tue May 27 2003 13:47:26.941342
>   last truncation:       Tue May 27 2003 13:47:26.941341
>   log head offset:          2614816
>   log tail offset:          2615588

Last truncation pretty much coincided with the last log record and we
only had about 772 bytes written to the log. My guess is that this was
nothing more than just a 'log truncated successfully' message.

>   space used by records:        772
>   space available:         12582140

Your original log size probably was 12MB.

> -----------------------------------------------
> After wiping the log with i /dev/hda7 38M (the size of the partition
> is 40MB):
...
>   space used by records:      33564
>   space available:         39812324

Now we have a 38MB log (38 * 1024 * 1024 = 39845888)

> I was then able to start up the codaserver and it seems to
> be working fine now. I'd be curious what went wrong, though - some 
> configuration mistake on my part?

No, my guess is that something went wrong during/after the truncation. A
write didn't hit the disk or something. Perhaps the server crashed
while it was committing the operation? Recovery sequentially steps
through the log and applies all modifications that are part of
successful transactions, whatever is left over is essentially
interpreted as an 'aborted' transaction.

Now that the server has managed to regain access to the RVM data and the
salvager has managed to recover all volumes we can be 99% sure that
everything is fine now. I have a server that went through a similar RVM
log badness and has been running without a problem for the past 2 years.
(where's that wood)

Jan