[Pacemaker] [Problem] The "dampen" parameter of the attrd_updater command is ignored, and an attribute is updated.

renayama19661014 at ybb.ne.jp renayama19661014 at ybb.ne.jp
Wed May 28 06:49:42 CEST 2014


Hi Andrew,

Thank you for comment.

> > --- attrd/command.c -----
> > (snip)
> >    /* this only involves cluster nodes. */
> >    if(v->nodeid == 0 && (v->is_remote == FALSE)) {
> >        if(crm_element_value_int(xml, F_ATTRD_HOST_ID, (int*)&v->nodeid) == 0) {
> >            /* Create the name/id association */
> >            crm_node_t *peer = crm_get_peer(v->nodeid, host);
> >            crm_trace("We know %s's node id now: %s", peer->uname, peer->uuid);
> >            if(election_state(writer) == election_won) {
> >                write_attributes(FALSE, TRUE);
> >                return;
> >            }
> >        }
> >    }
> 
> This is for 5194 right?

No.
I listed the same thing in 5194, but this does not seem to be 5194 basic problems.

I try the reproduction of 5194 problems, but have not been able to yet reappear.
Possibly 5194 problems may not happen in PM1.1.12-rc1.
 * As for 5194 matters, please give me time a little more.

> 
> I'd expect that block to hit this clause though:
> 
>      } else if(mainloop_timer_running(a->timer)) {
>         crm_info("Write out of '%s' delayed: timer is running", a->id);
>         return;

Which point of the source code does the suggested code mentioned above revise?
(Which line of the source code is it?)

Best Regards,
Hideo Yamauchi.

--- On Wed, 2014/5/28, Andrew Beekhof <andrew at beekhof.net> wrote:

> 
> On 27 May 2014, at 12:13 pm, renayama19661014 at ybb.ne.jp wrote:
> 
> > Hi All,
> > 
> > The attrd_updater command ignores the "dampen" parameter and updates an attribute.
> > 
> > Step1) Start one node.
> > [root at srv01 ~]# crm_mon -1 -Af
> > Last updated: Tue May 27 19:36:35 2014
> > Last change: Tue May 27 19:34:59 2014
> > Stack: corosync
> > Current DC: srv01 (3232238180) - partition WITHOUT quorum
> > Version: 1.1.11-f0f09b8
> > 1 Nodes configured
> > 0 Resources configured
> > 
> > 
> > Online: [ srv01 ]
> > 
> > 
> > Node Attributes:
> > * Node srv01:
> > 
> > Migration summary:
> > * Node srv01: 
> > 
> > Step2) Update an attribute by attrd_updater command.
> > [root at srv01 ~]# attrd_updater -n default_ping_set -U 500 -d 3000       
> > 
> > Step3) The attribute is updated without waiting for the time of the "dampen" parameter.
> > [root at srv01 ~]# cibadmin -Q | grep ping_set            
> >          <nvpair id="status-3232238180-default_ping_set" name="default_ping_set" value="500"/>
> > 
> > The next code seems to have a problem somehow or other.
> > 
> > --- attrd/command.c -----
> > (snip)
> >    /* this only involves cluster nodes. */
> >    if(v->nodeid == 0 && (v->is_remote == FALSE)) {
> >        if(crm_element_value_int(xml, F_ATTRD_HOST_ID, (int*)&v->nodeid) == 0) {
> >            /* Create the name/id association */
> >            crm_node_t *peer = crm_get_peer(v->nodeid, host);
> >            crm_trace("We know %s's node id now: %s", peer->uname, peer->uuid);
> >            if(election_state(writer) == election_won) {
> >                write_attributes(FALSE, TRUE);
> >                return;
> >            }
> >        }
> >    }
> 
> This is for 5194 right?
> 
> I'd expect that block to hit this clause though:
> 
>      } else if(mainloop_timer_running(a->timer)) {
>         crm_info("Write out of '%s' delayed: timer is running", a->id);
>         return;
> 
> 



More information about the Pacemaker mailing list