[ClusterLabs] Exception in adding resources to pacemaker

[Ken Gaillot]

On Thu, 2024-09-26 at 15:56 +0200, Miroslav Lisik wrote:
> On 9/26/24 08:11, bixiaoyan wrote:
> > After creating the group, there will be an error message when 
> > modifying the group meta attribute. The code snippet is as follows:
> > 
> > #/usr/sbin/pcs resource group add mygroupVGFS VIP PGSQL
> > 
> > #stdout:
> > 
> > #stderr:
> > 
> > #retcode:0
> > 
> > #/usr/sbin/pcs resource meta mygroupresource-stickiness=10
> > 
> > #stdout:
> > 
> > #stderr:Error:unable to find a resource/clone/group:mygroup
> > 
> > #retcode:1
> > 
> This is strage. After `pcs resource group add`, group should be in
> the
> CIB in case there is no error. Are these commands executed in
> sequence
> on the same node? Could you try to save the CIB before and after `pcs
> resource meta` command?
> 
> @Ken:
> Is there any chance that after executing `cibadmin --patch --verbose
> --xml-pipe`, execution of `cibadmin --local --query` will not return
> updated CIB?

I don't think so. cibadmin waits for a reply from the CIB manager, and
I believe the change is effective (at least locally) by that point.

> > 
> > 在 2024-09-25 21:13:46，"Miroslav Lisik" <mlisik at redhat.com> 写道：
> > > I'm forwarding this e-mail back to the users at clusterlabs.org
> > > mailing list.
> > > 
> > > On 9/25/24 10:26, 毕晓妍 wrote:
> > > > 
> > > > The version of pcs is 0.10.8, and resources are added by batch
> > > > calling 
> > > > the pcs command through a script. First, each resource is
> > > > created 
> > > > using pcs resource create, and then pcs resource group add is
> > > > executed 
> > > > to create the group. The resources in the group include LVM, 
> > > > Filesystem，IPaddr，pgsql。 The first 100+groups were added
> > > > normally 
> > > > through this method, but there were occasional exceptions
> > > > where 
> > > > resources could not be found. Is there a limit to the number
> > > > of 
> > > > resources added? Could it be that the large scale of XML has
> > > > caused 
> > > > this problem?
> > > > 
> > > 
> > > There is no limit on the number of resources added as Ken already
> > > wrote
> > > earlier in the thread.
> > > 
> > > Are you getting the error message 'unable to find a
> > > resource/clone/group' from the 'pcs resource group add' or
> > > different 
> > > command?
> > > 
> > > Could you provide a snippet with executed commands and their
> > > output?
> > > > 
> > > > 在 2024-09-25 01:56:39，"Miroslav Lisik" <mlisik at redhat.com> 写道：
> > > > > Hello,
> > > > > On 9/24/24 09:34, 毕晓妍 wrote:
> > > > > > 你好：
> > > > > > 我想请问一下pacemaker集群中添加的资源有数量上限吗？我在4节点集群中添加资源， 
> > > > > > 一组5个资源， 
> > > > > > 添加104组资源时，使用pcs命令添加没有任何报错，但是对组资源进行操作时显示unable 
> > > > > > to find a resource/clone/group。请问是什么原因导致的这一现象？
> > > > > > 
> > > > > > Hello,
> > > > > > 
> > > > > > I would like to inquire about the potential limit on the
> > > > > > quantity of 
> > > > > > resources that can be added to a pacemaker cluster.
> > > > > > Specifically, I've 
> > > > > > been working with a 4-node cluster configuration, initially
> > > > > > adding a 
> > > > > > single group comprising 5 resources. However, upon
> > > > > > attempting to 
> > > > > > augment the cluster by adding 104 additional resource
> > > > > > groups using the 
> > > > > > > pcs| command, I encountered an unexpected behavior: while
> > > > > > > the command 
> > > > > > execution itself did not report any errors, subsequent
> > > > > > operations 
> > > > > > targeting those groups resulted in a message stating
> > > > > > 'unable to find a 
> > > > > > resource/clone/group'.
> > > > > > 
> > > > > This error from pcs means, that the resource id used with a
> > > > > pcs command
> > > > > was not found in the CIB.
> > > > > 
> > > > > Either a resource was not created or a resource id was
> > > > > mistyped in a pcs
> > > > > command.
> > > > > 
> > > > > The `pcs resource create` should error in case resource
> > > > > cannot be
> > > > > created.
> > > > > 
> > > > > How do you run pcs commands? Manually or by some script?
> > > > > Could you
> > > > > provide more information like:
> > > > > * version of pcs
> > > > > * example of pcs commands used
> > > > > * cluster configuration (CIB, output of pcs config...)
> > > > > > Could you please shed some light on what might have caused
> > > > > > this 
> > > > > > phenomenon and whether there exists a known limitation with
> > > > > > respect to 
> > > > > > the total number of resources or resource groups that can
> > > > > > be 
> > > > > > effectively managed within a pacemaker cluster of this
> > > > > > size?
> > > > > > 
> > > > > > Looking forward to your reply.
> > > > > > 
> > > > > > 
> > > > > > _______________________________________________
> > > > > > Manage your subscription:
> > > > > > https://lists.clusterlabs.org/mailman/listinfo/users
> > > > > > 
> > > > > > ClusterLabs home: https://www.clusterlabs.org/
> > > > > Regards,
> > > > > Miroslav
> 
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users
> 
> ClusterLabs home: https://www.clusterlabs.org/
-- 
Ken Gaillot <kgaillot at redhat.com>