Re: Prometheus (OKD/3.11) NFS?

[Alan Christie <achristie informaticsmatters com>]

Thanks, it turned out to be relatively simple in that I just needed 2 PVs (prometheus-1 and 2) to satisfy prometheus and 3 (alertmanager-1 to 3) to satisfy the alertmanager. It wasn’t immediately obvious how to solve this in a ‘static’ system it is documented and now working - I just set the two `storage_enabled` variables but leave the `storageslass` ones alone.

I understand the side-effects of NFS but my options (at the moment) are extremely limited.

Thanks for your help.

Alan Christie
achristie informaticsmatters com



> On 12 Nov 2019, at 1:28 pm, Simon Pasquier <spasquie redhat com> wrote:
> 
> On Wed, Nov 6, 2019 at 6:11 PM Alan Christie
> <achristie informaticsmatters com> wrote:
>> 
>> Hi,
>> 
>> My cluster doesn’t have dynamic volumes at the moment. For experimentation is it possible to use NFS volumes in 3.11 for prometheus and the alert manager?
> 
> In general Prometheus doesn't play well with NFS because most
> implementations don't fully support POSIX.
> It is called out in the Prometheus documentation:
> https://prometheus.io/docs/prometheus/latest/storage/#operational-aspects
> 
>> 
>> I notice that the ansible playbook variables for prometheus in OKD 3.11 are very different to those in 3.7 but...
> 
> Yes the big difference starting with 3.11 is that Prometheus is
> deployed using the Prometheus operator via the Cluster Monitoring
> Operator.
> 
>> 
>> - Can I use pre-provisioned NFS volumes?
>> - …or…is there an equivalent of 3.7’s “openshift_prometheus_alertmanager_storage_kind=nfs"?
> 
> See https://docs.okd.io/3.11/install_config/prometheus_cluster_monitoring.html#persistent-storage
> 
> 
>> 
>> Any advice would be greatly appreciated.
>> 
>> Alan Christie
>> achristie informaticsmatters com
>> 
>> 
>> 
>> 
>> _______________________________________________
>> users mailing list
>> users lists openshift redhat com
>> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
>