I have installed a 3 node ES "Monitoringhttp://www.elasticsearch.org/guide/en/marvel/current/index.html"
cluster and configured two other ES clusters (10 nodes one an 3 nodes one)
to send data to it. The data gathering / storage seems to be working fine,
but the display using "latest" Marvel UI a bit confusing. For example, on
"Cluster Overview" page, how does one should pick a specific ES Cluster? At
the moment, the "Cluster Summary" is jumping between two ES cluster I have
seems to me randomly. I did try to add query criteria on top to make it
choose to no avail...
That said, the "Nodes" panel on the same page works as expected - i.e. by
defaults it show all reporting nodes (from both cluster), but if I specify
"Filter" there, I do get information about Nodes of a specific cluster.
I'm new to Marvel and above may seems a bit naive, but in its current form,
does "latest" Marvel was intended to support my Use Case? Thx.
At the moment marvel doesn't support monitoring multiple production
clusters with a single monitoring one. Working on it!
Cheers,
Boaz
On Friday, March 28, 2014 2:05:56 PM UTC+1, me wrote:
Hi there,
I have installed a 3 node ES "Monitoringhttp://www.elasticsearch.org/guide/en/marvel/current/index.html"
cluster and configured two other ES clusters (10 nodes one an 3 nodes one)
to send data to it. The data gathering / storage seems to be working fine,
but the display using "latest" Marvel UI a bit confusing. For example, on
"Cluster Overview" page, how does one should pick a specific ES Cluster? At
the moment, the "Cluster Summary" is jumping between two ES cluster I have
seems to me randomly. I did try to add query criteria on top to make it
choose to no avail...
That said, the "Nodes" panel on the same page works as expected - i.e. by
defaults it show all reporting nodes (from both cluster), but if I specify
"Filter" there, I do get information about Nodes of a specific cluster.
I'm new to Marvel and above may seems a bit naive, but in its current
form, does "latest" Marvel was intended to support my Use Case? Thx.
Looking forward to the update.
I presume this will also address the "Warning: Multiple masters. 2 nodes
report master role" of the nodes panel?
Perhaps it is possible to also use the cluster to differentiate between
indices?
Currently it seems like Marvel groups all data for indices based on its
name alone.
We have several clusters with the same index name (don't ask me why ),
which makes the index metrics useless since there is no distinction between
clusters.
On Wednesday, October 1, 2014 2:59:00 PM UTC+2, W Koot wrote:
Hi Boaz,
Looking forward to the update.
I presume this will also address the "Warning: Multiple masters. 2 nodes
report master role" of the nodes panel?
If you monitor two clusters, you will have multiple masters - which is for
a single cluster is a big no no. Multi cluster support in marvel will
obviously address this.
Perhaps it is possible to also use the cluster to differentiate between
indices?
Currently it seems like Marvel groups all data for indices based on its
name alone.
We have several clusters with the same index name (don't ask me why ),
which makes the index metrics useless since there is no distinction between
clusters.
We will use the cluster uuid to filter the data when displaying it. Same
name indices shouldn't be a problem.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.