On 10/06/2021 06:36, nina guo wrote:
Thank you Stuart.
For my understanding, the concept of HA is there are 2 instances -
master and slave. Master is running, and slave is standby. If master
breakes down, slave will take over.
But now for Prometheus, the 2 promethues instanaces are running at the
same time with scraping the same metrics.
I research the tool Promxy. It is mainly for aggregating the metrics.
It will deduplicate and merge the metrics.
I'm a little confuse about the HA on Promethues. It is a bit different
with the normal concept.
Yes. Prometheus does not run as a cluster or active/passive pair where
they communicate and transfer information between themselves. Instead
you run two totally separate Prometheus servers which maintain their own
(slightly different) view of the world and then query in a way that
prevents duplication.
Regarding alert manager, I noticed that it has native HA solution. So
that is to say, deploying 2 alert managers is OK? If there is any
alert, it will be sent 2 times with the 2 alert managers separately.
I'm still not very clear about the concept of HA for alert manager.
For Alertmanager the different instances communicate with each other
forming a cluster. Information about alerts is shared with a full copy
of that data held by each instance. Any Prometheus servers attached must
inform every member of an Alertmanager cluster. When alerts fire
Alertmanager will ensure a single notification is sent and will continue
to work even if cluster instances stop operating.
--
Stuart Clark
--
You received this message because you are subscribed to the Google Groups
"Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/prometheus-users/01d0dec8-c8b3-a551-62d2-8d2f6d3ab929%40Jahingo.com.
