[Pacemaker] [Question]The value of the attribute is changed by a timing and is sent.

[Andrew Beekhof]

On Tue, Aug 23, 2011 at 11:29 AM,  <renayama19661014 at ybb.ne.jp> wrote:
> Hi All,
>
> We confirmed the situation that a value from attrd was sent to by the change of the value just before that.
>
> (snip)
> Aug 14 20:05:33 srv01 attrd: [13880]: info: attrd_trigger_update: Sending flush op to all hosts for: default_ping_set (100)
> Aug 14 20:05:33 srv01 pingd: [14137]: info: stand_alone_ping: Node 192.168.100.1 is unreachable (read)
> Aug 14 20:05:33 srv01 attrd: [13880]: info: attrd_ha_callback: flush message from srv01
> Aug 14 20:05:33 srv01 attrd: [13880]: info: attrd_perform_update: Sent update 121: default_ping_set=0
> (snip)
>
> The movement that we supposed from log is as follows.
>
> Step1) The timer of the value of dampen(default=5s) worked.
>          Attrd transmits flush message to transmit default_ping_set attribute 100.
> Step2) However, the communication of the network fell just after that.
>          The default_ping_set level in attrd is changed to 0 by pingd.
> Step3) Attrd was going to transmit 100 at the point in time when I transmitted flush message(Step1), but attrd in response to flush message transmitted a value of default_ping_set with 0.
>
> Is this movement specifications?

I can see how this might be undesirable, but it is in line with how
I'd expect attrd to behave.

> Or is it a bug?
>
> If attrd transmits a latest attribute level by all means, I can understand this movement when it is specifications.

Right. The flush message means "write whatever value you have right now".

>
> Best Regards,
> Hideo Yamauchi.
>
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>