[Opendnssec-user] ods-signer SEGV on zone deletion

Havard Eidnes he at uninett.no
Thu Dec 17 09:18:01 UTC 2015


my struggles with OpenDNSSEC continues.  I recently had to delete
9 zones from our OpenDNSSEC installation.

I did this via

  ods-ksmutil zone delete --zone <zonename>

It worked OK for the first 8 zones, but for the last one I got
"connection refused", and in the kernel log (where I've already
turned on logging of abnormal signal-triggered exits), I found:

pid 335 (ods-signerd), uid 1072: exited on signal 6 (core not dumped, err = 1)

Signal 6 is SIGSEGV, but annoyingly I didn't get a core dump, err = 1
is EPERM.  I don't quite understand this -- our currently-running
ods-signerd has /var/opendnssec/tmp as current working directory:

hugin# ps ax | grep ods-signerd
10447 ?      Isl     4:01.28 /usr/pkg/sbin/ods-signerd -c /usr/pkg/etc/opendnss
 1852 pts/15 R+      0:00.00 grep ods-signerd 
hugin# fstat -p 10447
USER     CMD          PID   FD MOUNT       INUM MODE         SZ|DV R/W
ods      ods-signerd 10447   wd /var     6447361 drwxr-xr-x   31744 r 
hugin# find /var -inum 6447361
hugin# id 1072
uid=1072(ods) gid=1072(ods) groups=1072(ods)
hugin# ls -ld /var/opendnssec/tmp
drwxr-xr-x  2 ods  ods  31744 Dec 17 10:10 /var/opendnssec/tmp

So ... I don't understand why I could not get a core dump to
examine and provide more information.  I don't suppose the
signerd "hops around" with its current working directory?

So this is just a "this can apparently happen" message.

And shortly after startup of a new signer, I of course hit the dreaded
and already reported "soamin not set" assertion, so had to once again
manually tidy up and re-start the signer and signal re-transfer of the
zones from the hidden master via notify messages.


- Håvard
Opendnssec-user mailing list
Opendnssec-user at lists.opendnssec.org

More information about the Opendnssec-user mailing list