[Opendnssec-user] ods-signerd thread abnormal running

[yaohongyuan]

Hi all,
We use the opendnssec version is 1.4.7 .
Yesterday I noticed that ods-signerd thread work unsteadiness in our test env , if run after 8 hours this thread will disappear ... and get system log as below :
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test2 cannot tcp write to 202.173.9.81: Broken pipe
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test17 received too short udp reply from 202.173.9.81, retry tcp
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test17 request tcp/ixfr=2015101431 to 202.173.9.81
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test16 received too short udp reply from 202.173.9.81, retry tcp
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test16 request tcp/ixfr=2015101647 to 202.173.9.81
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: wire/notify.c:476: notify_handle_zone: assertion notify->handler.fd == -1 failed

And I added a crontab job to monitor if the ods-signerd thread disappeared then run the ./ods-signerd shell . But in above case it's not work , got some system logs as below :
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test2 cannot tcp write to 202.173.9.81: Broken pipe
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test17 received too short udp reply from 202.173.9.81, retry tcp
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test17 request tcp/ixfr=2015101431 to 202.173.9.81
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test16 received too short udp reply from 202.173.9.81, retry tcp
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: [xfrd] zone test16 request tcp/ixfr=2015101647 to 202.173.9.81
Mar 23 22:52:04 p01-test-devops-9-81 ods-signerd: wire/notify.c:476: notify_handle_zone: assertion notify->handler.fd == -1 failed
Mar 23 22:53:01 p01-test-devops-9-81 ods-signerd: [hsm] hsm_get_slot_id(): No slots found in HSM
Mar 23 22:53:01 p01-test-devops-9-81 ods-signerd: [engine] setup failed: HSM error
Mar 23 22:53:01 p01-test-devops-9-81 ods-signerd: [engine] signer shutdown


I have to close the ods-enforcerd thread then start the ods-signerd will back to normal .
Dose anybody know what cause the ods-signerd thread disappear ? And when I restart the ods-signerd why can't found slots in HSM ? (I've checked the HSM status is OK.)
Thank you very much.
With kind regards,
Dean

[Yuri Schaeffer]

Hi Dean,

Post by yaohongyuan
We use the opendnssec version is 1.4.7 .
Yesterday I noticed that ods-signerd thread work unsteadiness in
our test env , if run after 8 hours this thread will disappear ... and

We fixed a couple of bugs in the last few months regarding unexpected
TCP packets. I advice you to upgrade to version 1.4.9. That might very
well fix it.

Post by yaohongyuan
I have to close the ods-enforcerd thread then start
the ods-signerd will back to normal .
Dose anybody know what cause the ods-signerd thread disappear ?
And when I restart the ods-signerd why can't found slots in HSM ? (I've
checked the HSM status is OK.)

I have no answer for this at this point.

//Yuri