nsd users - Aug 2009 - corrupted database

Aug 18 17:13:01 gir nsd[2744]: could not read ixfr store info: file format error
Aug 18 17:13:01 gir nsd[2744]: error processing diff file
Aug 18 17:13:01 gir nsd[2744]: unable to load the diff file: /var/nsd/ixfr.db
Aug 18 17:13:01 gir nsd[26056]: Reload process 2744 failed with status 256,
continuing with old database

I had a server crash, brought it back up, and ran into these problems.
Hand deleting, a restart and rebuild, fixed it.

Would it make sense for nsd to totally rebuild the diff file (and
any other 'temporary' files) in such a case?

I am running nsd 3.2.2 on OpenBSD.

-- 
I cannot conceive that anybody will require multiplications at the rate
of 40,000 or even 4,000 per hour ...
		-- F. H. Wales (1936)

On Tue, 18 Aug 2009, Peter Hessler wrote:
> Aug 18 17:13:01 gir nsd[2744]: could not read ixfr store info: file format
error
> Aug 18 17:13:01 gir nsd[2744]: error processing diff file
> Aug 18 17:13:01 gir nsd[2744]: unable to load the diff file:
/var/nsd/ixfr.db
> Aug 18 17:13:01 gir nsd[26056]: Reload process 2744 failed with status 256,
continuing with old database
>
> I had a server crash, brought it back up, and ran into these problems.
> Hand deleting, a restart and rebuild, fixed it.
> Would it make sense for nsd to totally rebuild the diff file (and
> any other 'temporary' files) in such a case?
You cannot "rebuild" the diff file. Its the changes your received via
axfr/notify.
You need to regularly run "nsd patch" to write the diffs back into
their zone files.

Then you could automate removing corrupted diffs, though one might still prefer
to know this over automatically fixing and perhaps losing changes to the zones.

Paul