-
Type:
Story
-
Status: Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: 1.3.9, 1.3.10
-
Fix Version/s: None
-
Component/s: Signer
-
Labels:None
-
Environment:
Debian/Squeeze on OpenVZ with ODS 1.3.10 (weekly)
RHEL 6.3 Santiago with ODS 1.3.9 (once, probably the same issue)
The signer gets stuck about once a week, and then shows a queue like
root@dnsmaxi:/# ods-signer queue
It is now Mon Nov 26 10:14:38 2012
I have 54 tasks scheduled.
On Sun Nov 25 08:22:57 2012 I will [sign] zone A
On Sun Nov 25 08:22:57 2012 I will [sign] zone B
On Sun Nov 25 08:22:57 2012 I will [sign] zone C
On Sun Nov 25 08:22:57 2012 I will [sign] zone D
On Sun Nov 25 08:22:57 2012 I will [sign] zone E
On Sun Nov 25 08:22:58 2012 I will [sign] zone F
It suffices to restart the signer, and even an explicit "ods-signer sign X" seems to do the trick (X need not be A). It looks like a race condition, but since it occurs often on my platform I could collect information when it happens again. What information would be useful? A core dump, a queue dump like above, more...? If it happens I want to continue signing as soon as possible, but I can prepare to collect things as they are needed.