<div dir="ltr">Hi Ken,<div><br></div><div>I didn't look into the logs, but I experimented with it for a while.</div><div>Here is what I found.</div><div><br></div><div>It worked for you because this attribute - "my-attr" - has not ever been set before in that cluster.</div><div><br></div><div>So if you set an attribute, then remove it, and then set it with "--delay", like:</div><div><br></div><div> # attrd_updater -N node-0 -n my-attr --update false --delay 20</div><div><br></div><div>, this delay (dampening) won't work.</div><div>Moreover, when you delete this attribute the actual remove will be delayed by that "--delay" which was used when the attribute was set.</div><div><br></div></div><div class="gmail_extra"><br clear="all"><div><div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr">Thank you,<div>Kostia</div></div></div></div></div></div>
<br><div class="gmail_quote">On Tue, Nov 29, 2016 at 1:08 AM, Ken Gaillot <span dir="ltr"><<a href="mailto:kgaillot@redhat.com" target="_blank">kgaillot@redhat.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On 11/24/2016 05:24 AM, Kostiantyn Ponomarenko wrote:<br>
> Attribute dampening doesn't work for me also.<br>
> To test that I have a script:<br>
><br>
> attrd_updater -N node-0 -n my-attr --update false --delay 20<br>
> sleep 3<br>
> attrd_updater -N node-0 -n my-attr --update true<br>
> sleep 7<br>
> attrd_updater -N node-1 -n my-attr --update true<br>
<br>
</span>This sequence works for me -- the attributes are not written to the live<br>
CIB until the end of the delay, when both are written at the same time.<br>
<br>
The remaining issue must be with the policy engine. You could look at<br>
the detail log on the DC when these changes were made; you should see<br>
info-level messages with the CIB change with both values together (lines<br>
with "cib_perform_op: ++" and the attribute values), then "Transition<br>
aborted" with "Transient attribute change", then a bunch of "pengine:"<br>
lines saying what the cluster wants to do with each resource.<br>
<br>
There should be some information about the scores used to place the<br>
resources.<br>
<span class=""><br>
><br>
> All my resources have this rule in Pacemaker config:<br>
><br>
> crm configure location res1-location-rule res1 \<br>
> rule 0: my-attr eq true \<br>
> rule -inf: my-attr ne true<br>
><br>
> On a working two-node cluster I remove "my-attr" from both nodes.<br>
> Then run my script. And all resources start on node-0.<br>
> Am I doing something wrong?<br>
> Or maybe my understanding of an attribute dampening is not correct?<br>
><br>
> My Pacemaker version is 1.1.13. (heh, not the last one, but it is what<br>
> it is ...)<br>
><br>
> Thank you,<br>
> Kostia<br>
><br>
> On Wed, Nov 23, 2016 at 7:27 PM, Kostiantyn Ponomarenko<br>
> <<a href="mailto:konstantin.ponomarenko@gmail.com">konstantin.ponomarenko@gmail.<wbr>com</a><br>
</span><span class="">> <mailto:<a href="mailto:konstantin.ponomarenko@gmail.com">konstantin.<wbr>ponomarenko@gmail.com</a>>> wrote:<br>
><br>
> Maybe I am doing something wrong, but I cannot set "status" section<br>
> node attributes to a shadow cib, cluster applies them immediately.<br>
> To try it out I do in a console:<br>
><br>
> crm_shadow --create test<br>
> crm_attribute --type nodes --node node-0 --name my-attribute<br>
> --update 1 --lifetime=reboot<br>
><br>
> And this attribute is set to the live cluster configuration immediately.<br>
> What am I doing wrong?<br>
><br>
> Thank you,<br>
> Kostia<br>
><br>
> On Tue, Nov 22, 2016 at 11:33 PM, Kostiantyn Ponomarenko<br>
> <<a href="mailto:konstantin.ponomarenko@gmail.com">konstantin.ponomarenko@gmail.<wbr>com</a><br>
</span><span class="">> <mailto:<a href="mailto:konstantin.ponomarenko@gmail.com">konstantin.<wbr>ponomarenko@gmail.com</a>>> wrote:<br>
><br>
> Ken,<br>
> Thank you for the explanation.<br>
> I will try this low-level way of shadow cib creation tomorrow.<br>
> PS: I will sleep much better with this excellent news/idea. =)<br>
><br>
> Thank you,<br>
> Kostia<br>
><br>
> On Tue, Nov 22, 2016 at 10:53 PM, Ken Gaillot<br>
</span><span class="">> <<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a> <mailto:<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a>>> wrote:<br>
><br>
> On 11/22/2016 04:39 AM, Kostiantyn Ponomarenko wrote:<br>
> > Using "shadow cib" in crmsh looks like a good idea, but it doesn't work<br>
> > with node attributes set into "status" section of Pacemaker config.<br>
> > I wonder it it is possible to make it work that way.<br>
><br>
> Forgot to mention -- the shadow CIB is probably the best way<br>
> to do this.<br>
> I don't know if there's a way to do it in crmsh, but you can<br>
> use it with<br>
> the low-level commands crm_shadow and crm_attribute<br>
> --lifetime=reboot.<br>
><br>
> > Ken,<br>
> >>> start dampening timer<br>
> > Could you please elaborate more on this. I don't get how I can set this<br>
> > timer.<br>
> > Do I need to set this timer for each node?<br>
> ><br>
> ><br>
> > Thank you,<br>
> > Kostia<br>
> ><br>
> > On Mon, Nov 21, 2016 at 9:30 AM, Ulrich Windl<br>
> > <<a href="mailto:Ulrich.Windl@rz.uni-regensburg.de">Ulrich.Windl@rz.uni-<wbr>regensburg.de</a><br>
> <mailto:<a href="mailto:Ulrich.Windl@rz.uni-regensburg.de">Ulrich.Windl@rz.uni-<wbr>regensburg.de</a>><br>
</span>> > <mailto:<a href="mailto:Ulrich.Windl@rz.uni-regensburg.de">Ulrich.Windl@rz.uni-<wbr>regensburg.de</a><br>
<span class="">> <mailto:<a href="mailto:Ulrich.Windl@rz.uni-regensburg.de">Ulrich.Windl@rz.uni-<wbr>regensburg.de</a>>>> wrote:<br>
> ><br>
> > >>> Ken Gaillot <<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a><br>
</span>> <mailto:<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a>> <mailto:<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a><br>
<span class="">> <mailto:<a href="mailto:kgaillot@redhat.com">kgaillot@redhat.com</a>>>><br>
> > schrieb am 18.11.2016 um 16:17 in Nachricht<br>
> > <<a href="mailto:d6f449da-64f8-12ad-00be-e772d8e382ca@redhat.com">d6f449da-64f8-12ad-00be-<wbr>e772d8e382ca@redhat.com</a><br>
> <mailto:<a href="mailto:d6f449da-64f8-12ad-00be-e772d8e382ca@redhat.com">d6f449da-64f8-12ad-<wbr>00be-e772d8e382ca@redhat.com</a>><br>
> ><br>
</span>> <mailto:<a href="mailto:d6f449da-64f8-12ad-00be-e772d8e382ca@redhat.com">d6f449da-64f8-12ad-<wbr>00be-e772d8e382ca@redhat.com</a><br>
<div class="HOEnZb"><div class="h5">> <mailto:<a href="mailto:d6f449da-64f8-12ad-00be-e772d8e382ca@redhat.com">d6f449da-64f8-12ad-<wbr>00be-e772d8e382ca@redhat.com</a>>><wbr>>:<br>
> > > On 11/18/2016 08:55 AM, Kostiantyn Ponomarenko wrote:<br>
> > >> Hi folks,<br>
> > >><br>
> > >> Is there a way to set a node attribute to the<br>
> "status" section for few<br>
> > >> nodes at the same time?<br>
> > >><br>
> > >> In my case there is a node attribute which allows<br>
> some resources to<br>
> > >> start in the cluster if it is set.<br>
> > >> If I set this node attribute for say two nodes in a<br>
> way - one and then<br>
> > >> another, than these resources are not distributed<br>
> equally between these<br>
> > >> two nodes. That because Pacemaker picks the first<br>
> node to with this<br>
> > >> attribute is set and immediately starts all allowed<br>
> resources on it. And<br>
> > >> this is not the behavior i would like to get.<br>
> > >><br>
> > >> Thank you,<br>
> > >> Kostia<br>
> > ><br>
> > > Not that I know of, but it would be a good feature<br>
> to add to<br>
> > > attrd_updater and/or crm_attribute.<br>
> ><br>
> > With crm (shell) you don't have transactions for node<br>
> attributes,<br>
> > but for the configuration. So if you add a location<br>
> restriction<br>
> > preventing any resources on your nodes, then enable<br>
> the nodes, and<br>
> > then delete the location restrictions in one<br>
> transaction, you might<br>
> > get what you want. It's not elegant, but itt ill do.<br>
> ><br>
> > To the crm shell maintainer: Is is difficult to build<br>
> transactions<br>
> > to node status changes? The problem I see is this: For<br>
> configuration<br>
> > you always have transactions (requiring "commit), but<br>
> for nodes you<br>
> > traditionally have non (effects are immediate). So<br>
> you'd need a<br>
> > thing like "start transaction" which requires a<br>
> "commit" or some<br>
> > kind of abort later.<br>
> ><br>
> > I also don't know whether a "shadow CIB" would help<br>
> for the original<br>
> > problem.<br>
> ><br>
> > Ulrich<br>
> ><br>
> > ><br>
> > > You can probably hack it with a dampening value of a<br>
> few seconds. If<br>
> > > your rule checks for a particular value of the<br>
> attribute, set all the<br>
> > > nodes to a different value first, which will write<br>
> that value and<br>
> > start<br>
> > > the dampening timer. Then set all the attributes to<br>
> the desired value,<br>
> > > and they will get written out together when the<br>
> timer expires.<br>
</div></div></blockquote></div><br></div>