[ClusterLabs] Integrate external alerts with pcs cluster

[Ajay Srivastava]

Hi Strahil,Yes. It's simple but monitoring is done by an external service which sends alarm in case state of a hardware changes. It's a complex code and multiple devices are being monitored by it. If I ask pacemaker to monitor the device, this code has to be replicated in resource agent's monitor which is not an option. Instead of that, I plan to examine the alert in external service and fail the resource if alarm is critical. Hope that it clarifies the scenario.By the way, how could I fail the resource from outside. I see that option in crm_resource CLI (--fail) but not in pcs CLI.Regards,AjayFrom: Strahil Nikolov <hunter86_bg at yahoo.com>Sent: Sat, 11 Apr 2020 18:50:25To: Cluster Labs - All topics related to open-source clustering welcomed <users at clusterlabs.org>,Ajay Srivastava <ajay_srivastava at rediffmail.com>Subject: Re: [ClusterLabs] Integrate external alerts with pcs clusterOn April 11, 2020 2:04:23 PM GMT+03:00, Ajay Srivastava <ajay_srivastava at rediffmail.com> wrote:>Hi,In my environment I have a pacemaker cluster running which has>various software services as resource agents.These services have>dependency on some hardware components. There is a service which>monitors the hardware and sends alerts if anything is wrong with the>hardware.My plan is to add a resource agent for each hardware component>with proper dependencies and fail this resource agent if I get a>critical alert from hardware monitoring service. Please note that I do>not want to use monitor functionality of resource agent as there is a>service which is already doing same thing. I have two queries here -1)>Does the approach look good ? Is there a better way to implement it in>pacemaker cluster ?2) I can find --fail option in crm_resource but not>in pcs cli. What would be the equivalent command in pcs as I am using>pcs cli to configure the cluster ?Regards,Ajay&nbsp;Hi Ajay,Sadly I don't get the logic.Based  on your  e-mail you want to create a cluster  resource that will fail based on hardware issue.Yet, are you going to put  other cluster resources as a dependency (for example  -> ordering constraint that stops some  application in case some cpu has  stuck for  '120s') ?All the cluster  needs  is to know how to:- Start the  resource- Monitor  the liveliness  of the resource- Stop the resourceBest Regards,Strahil Nikolov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clusterlabs.org/pipermail/users/attachments/20200411/3806a746/attachment.htm>