[Opendnssec-user] softhsm corruption

Jaap Akkerhuis jaap at NLnetLabs.nl
Fri Feb 20 20:27:46 UTC 2015


 Paul Wouters writes:

 > On Fri, 20 Feb 2015, Patrik Wallström wrote:
 > 
 > >> I'm not sure what caused this. The disk was not full and still has 1Gb
 > >> of free space.
 > >
 > > You didn’t have a backup?
 > 
 > I do, but sadly it does not contain all the current ZSKs. Not the end of
 > the world. I fixed the backup to also pull in /var/softhsm every night
 > now :P
 > 
 > > The files are sqlite3 database files. Searching the net on how to recover from a corrupt sqlite3 database gives you advice such as it might be corrupted indices, but the content might be restored in a new database using
 > >
 > > sqlite3 slot0.db ".dump” | sqlite3 newslot0.db
 > 
 > It looks like the slot0.db file on disk is 6kb while the backup I
 > restored was around 325kb. So I don't think there is anything left
 > to recover. I'll try and find the relevant log entries to see if
 > we might see why this happened.

A lot of .db have "holes" which are replaced with a lot of null
bytes when you copy them. Maybe that is what you are seeing?

	jaap



More information about the Opendnssec-user mailing list