Hello Anil - thanks for your reply
Im glad that someone else is experiencing this too!
Ive made my own custom monitor.. using vb.net
- its simply an application that runs every so many minutes (currently 5)
via scheduled task on windos machine that looks at the health of the
cluster..
so it would look at the ipaddress where the elastic db is then the port
number..
so for example http://192.169.1.1:9200
then the api that looks for the health.. so the entire url would be.
http://192.169.1.1:9200/_cluster/health?pretty=true
and I get the below result...
{
"cluster_name" : "elasticsearch",
"status" : "green",
"timed_out" : false,
"number_of_nodes" : 2,
"number_of_data_nodes" : 2,
"active_primary_shards" : 11,
"active_shards" : 22,
"relocating_shards" : 0,
"initializing_shards" : 0,
"unassigned_shards" : 0
}
.. then the bit that says status green..
I look for that in my app and if its anything else other than green, I make the app email myself.
This works for me, BUT.. it still does not explain WHY these things happen, and I would rather diagnose why these happen,to stop them from happening again.
I dont have nodes leaving my cluster..
I have two separate machines running windows, and they are both always in the cluster, but for whatever reason - the cluster sees fit to turn yellow on occasion, (not regularly, but every now and then) which I understand that replicas didnt happen on some shards.. but I want to know whyyyyyy lol
Cheers
On Tuesday, 3 March 2015 12:33:10 UTC, Anil Karaka wrote:
What sort of monitoring are you having.. and how are you getting alerts
when the state of the cluster changes..? Is there some plugin that I can
install?
We are facing similar problems as well.. Even more scary my nodes are
leaving the cluster from time to time and rejoin the cluster.. Because of
this lots of shards go to "unassigned" state and it takes several minutes
for the cluster to get back to normal green state..
I came across this blog post that suggested me to change my garbage
collector to G1GC from the default ConcMarkSweepGC, so that stop the world
garbage collections will not occur anymore....
Still my nodes are leaving the cluster
Can you tell me what sort of monitoring are you using to get alerts when
cluster state changes.
Thank you.
On Tuesday, March 3, 2015 at 4:19:13 PM UTC+5:30, Steven Williams wrote:
Hello all
I have 2 nodes (two different machines) and on occasion the cluster goes
from green to yellow (my cluster health is below, and at the moment all is
good)
"cluster_name" : "elasticsearch",
"status" : "green",
"timed_out" : false,
"number_of_nodes" : 2,
"number_of_data_nodes" : 2,
"active_primary_shards" : 11,
"active_shards" : 22,
"relocating_shards" : 0,
"initializing_shards" : 0,
"unassigned_shards" : 0
}
I have read what the different status colours mean, but I cant find
anywhere that says WHY these things happen i.e. network issue, cpu load etc
etc.
Are the any main reasons that would give cause to red / yellow health?
ive got monitoring on my cluster to alert me when its not green, and I
can go and restart the slave server, then it goes back to green.. but still
this does not identify the root cause as to why this happens?
Any comments would be greatly appreciated
Cheers
--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/b34b57eb-427a-4e6d-b536-91cad205f438%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.