On Jan 25, 2008, at 3:48 AM, HIDEO YAMAUCHI wrote:
Hi,
I examined it about the reboot of the attrd process of Heartbeat.
1)I update an attribute with a d option by an attrd_updater command.
attrd_updater -n ... -v ... -d 10
2)I kill of an attrd process immediately.
kill -9 ATTRD_PID
3)The attrd process is rebooted
4)But, the instructions of the attrd_updater command are not carried
out.
The same problem occurs by the notice from pingd. ( example :
pingd ... -d 5s )
I think that the application of the suicide code is necessary for
attrd to break off this problem.
Will you update attrd to become failfast?
Thats a very big hammer you're proposing to use there.
What's the real-world example of 2) happening? If some hacker gets in
as root, the least of your problems is them killing the attrd
process :-)
A smarter solution might be to simply have attrd clients wait to be
notified that the update was successful (attrd exiting before this
notification would be considered a NACK).
This way they, or their caller, can simply try again.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
