[Opendnssec-user] signer process hangs after finishing the work

Michael Braunoeder mib at nic.at
Mon Dec 7 10:18:38 UTC 2009

Hi Rickard,

>> This issue has been resolved in trunk yesterday. And it will be included in the next release. The problem was that the zone was scheduled for signing, because the Signer Engine did not "remember" that a signing of the zone was in progress. Thus getting multiple threads working on the same zone.
> Perfect :-)
> I'm doing the upgrade and test it.

Unfortunately this issue seems not to be solved completly, I did an 
upgrade to the last SVN-trunk last Friday (Revision 2571), removed all 
input, tmp and output files and restarted the system. Currently (after 
an automatic run over the weekend) there are about 20 old signer 
processes running.

I also noticed a 2nd issue: the zone_fetcher writes sometimes an 
incomplete zonefile. It looks like that sometimes the last buffer isn't 
written to the zonefile. The retrieved zonefile looks like this

z-domain.or.at.     10800   IN      NS      ns.domain.at.
z-domain.or.at.     10800   IN      NS      ns2.domain.at.

and the last about 20 domains are missing and the sorter complains about 
Syntax errors in the zone. When I remove the broken zonefile and send a 
new notify to the zone_fetcher, the zones are retrieved completly. But 
after a few runs again a broken zonefile is produced. This issus occurs
only on 2 small 2nd-level domains (zonefile size about 3,2MB and 730KB). 
The logfiles look good (no errormessages), if needed I can send you a copy.


More information about the Opendnssec-user mailing list