[ClusterLabs] Antw: [EXT] Colocation per site ?

[Strahil Nikolov]

Actually , I found the error with the help of the RH support.
I was colocating with the Master SAPHanaController resource and thus the cluster did what I have told it to do ;)
Now, I just colocated (node-attribute=hana_sid_site) with the first backup ip  (which is colocated with the master) and added a location constraint for bkpip2 & bkpip3 and works like a charm.
I am still considering wether to keep the delay primitive that prevents the backup IPs to be brought up immediately.
The problem is that the current backup solution will be triggered from each HANA node , and as we got a lot of custom stuff - I want to make it fool-proof as much as possible. I've already organised a discussion about those backup IPs.

Best Regards,Strahil Nikolov
 
 
  On Wed, Mar 31, 2021 at 10:54, Andrei Borzenkov<arvidjaar at gmail.com> wrote:   On Wed, Mar 31, 2021 at 8:34 AM Strahil Nikolov <hunter86_bg at yahoo.com> wrote:
>
> Damn... I am too hasty.
>
> It seems that the 2 resources I have already configured are also running on the master.
>
> The colocation constraint is like:
>
> rsc_bkpip3_SAPHana_SID_HDBinst_num with rsc_SAPHana_SID_HDBinst_num-clone (score: INFINITY) (node-attribute:hana_sid_site) (rsc-role:Started) (with-rsc-role:Master)
>
>
> Setting location constraints to prevent from running on the masters causes the resource to stop.
>

Well, this constraint does not prevent two resources from running on
the same node. You would need to either use clone and restrict the
number of total and per-node clone instances or use negative
colocation between each pair of your resources (so they are not
started on the same node).

But I suspect you really overengineer it. At the end SAP HANA backup
is always initiated by SAP HANA itself and you only need the single IP
address to do it. This single IP address is already available as part
of SAP HANA replication setup (and should be active on the master HANA
node). Backup software usually wants to know which *physical* servers
a given virtual client may be allowed to run on, but in this case it
should really be enough to simply add all nodes to this list.

You may want to discuss it with your backup team again.
  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20210331/feadf4b1/attachment-0001.htm>