Yup thanks , that's what I thought.
On 22/04/2015 2:49 pm, David Pilato da...@pilato.fr wrote:
Only post 1.0
--
David ;-)
Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
Le 22 avr. 2015 à 01:14, Norberto Meijome num...@gmail.com a écrit :
David, is this the case with older
David, is this the case with older versions (both client and server on
0.90.x versions using java client), and across the 0.90 to 1.x boundary, or
only post 1.x?
On 22/04/2015 12:03 am, David Pilato da...@pilato.fr wrote:
This should work in both ways.
The client knows what is the node
Replica = 3 means 4 copies of your data ( for each shard, 1 master and 3
replicas)
On 21/04/2015 7:54 am, TB txind...@gmail.com wrote:
I have my indexes size @ 6 GB currently with replica set @ 1.
I have 3 node cluster, in order to utilize the cluster , my understanding
that i would have set
Hi,
My gut feel is don't add this to the ES setup itself. Horses for courses -
have your script (Python +1) running somewhere taking care of the
processing, dealing with issues on the ftp side , etc. Let ES do its
thing...specially if the XML parsing will take so much memory and you need
external
Replica =1 means you have 1 replica of the master shards - I.e. you have 2
copies of the data in total.
On 02/04/2015 2:06 pm, Nishad Karekar nishad.kare...@acxiom.com wrote:
I am unable to understand the results from the CAT API
curl
yes, that's the difference between a (network service which exposes ) block
storage like iSCSI and a network file system like NFS ( or glusterFS or
Lustre... ).
I don't see why on a local device (iSCSI) you'd have any issue with numeric
uid not matching the 'name' of the user - unless, of course,
TCP/9200 is for REST interface...Zen ping should be on 9300 ... I suspect
you set a config wrong...
On 11/03/2015 4:33 pm, Monika Bhadauria mon...@codeignition.co wrote:
Hi guys,
I have my Elasticsearch on one server and logstash on another.
I am getting the following error in my logstash,
?
Tagging nodes probably won't help in this instance.
Basically if you want to shut everything down you need to go through
recovery, and depending on how long that takes it may not be worth the
cost. This is something you need to test.
On 25 February 2015 at 18:14, Norberto Meijome num...@gmail.com
, it just blows fully to 395% (~4*100%).
On Friday, February 20, 2015 at 1:20:16 PM UTC+1, Norberto Meijome wrote:
Hi Maik,
Have you tried changing bulk size? May also be worth seeing if
separating masters to their own nodes makes a difference...
On 20/02/2015 8:22 pm, Maik Broxterman broxt
BTW, are you reducing / disabling the refresh rate while bulk indexing ?
On 22/02/2015 10:08 am, Norberto Meijome num...@gmail.com wrote:
OK, so what you have is resource contention between searches and
indexing...
On 22/02/2015 12:44 am, Maik Broxterman broxter...@gmail.com wrote:
Ok, I
Hi Maik,
Have you tried changing bulk size? May also be worth seeing if separating
masters to their own nodes makes a difference...
On 20/02/2015 8:22 pm, Maik Broxterman broxter...@gmail.com wrote:
Hello,
We are currently in the process of moving from an ES 0.9 cluster to an ES
1.4 cluster.
Sure...the interesting point in the OP is the fact both servers are in
different VPCs - not sure if it should be possible to resolve across vpcs
...
On 08/02/2015 7:04 pm, Ivan G igonzalezvalie...@gmail.com wrote:
DNS queries inside vpc are resolved to the internal IP by aws servers.
One
Are you referring to the master server by fqdn or IP? If fqdn, don't
forget about ec2's split horizon (though I don't think it should be
resolvable across 2 separate VPCs...
Can you open a socket from host 1 to host 2 manually (with nc or telnet) on
TCP/9300?
On 08/02/2015 9:43 am, Eugen
The load issue affecting master detection / election shouldn't happen if
you have dedicated masters... At least it is with 0.90.x
( with my limited knowledge of ES implementation details, there seems to be
a lock or priority issue when serving large # of requests (http / thrift) ,
affecting
Yes..but this might not be an option if your instance is in a private
subnet...it also means handling all your IPS like this ( though in theory
you don't need internal IPs, security group id/name would do as well...) -
there r limits to how many rules you can add to a secgroup
At the same
Hi David,
Indeed, the plugin makes AWS API calls ( ec2 describe instances) in order
to find candidates to cluster with. Unfortunately, if memory serves me
right, those are to external IPs...
Hint - tinyproxy with whitelist on your nat gw , and proper env
configuration so that the client side
FWIW, we saw many long running GC events using the default GC manager -
changing to G1 solved most of the problems ( at the expense of slightly
higher CPU all the time) After that you can take the longer road to
debugging memory allocation for your use case :-)
On 18/11/2014 6:21 am, Wilfred
I am pretty sure you can open the ports for the sec group the elb belongs
to , regardless of the az. (Az, not region). Unless you r using network
acls.
Anyway, not really ES... pm me if u want to continue the AWS discussion :-)
On 16/10/2014 3:37 pm, Zoran Jeremic zoran.jere...@gmail.com wrote:
Or, if your use case allows for it, have a very well oiled rebuild process
(data included).
On 14/10/2014 8:36 am, Itamar Syn-Hershko ita...@code972.com wrote:
Yes, you don't want to use anything other than local storage for
Elasticsearch. Not EBS and definitely not S3. You can use the
, Norberto Meijome wrote:
Zoran, good to hear it is working now.
It should work pretty well with ec2 auto discovery - unicast is a good
starting point but unless you are statically assigning them via cloud
formation (or manually?), it may not be worth the trouble (and it stops you
from dynamically
Zoran, good to hear it is working now.
It should work pretty well with ec2 auto discovery - unicast is a good
starting point but unless you are statically assigning them via cloud
formation (or manually?), it may not be worth the trouble (and it stops you
from dynamically scaling your cluster)
-
Hi, not sure tbh
Kibana is a js interface so I don't think it makes sense to alert from it.
You could monitor the results stored in ES with nagios/zabbix/ your
monitoring of choice and parse the json result and alert based on that.
We've used logstash's statsd module to send data we are interested
@mauri, thank you for such interesting analysis.
On 21/03/2014 1:01 PM, Mauri ma...@proactive-edge.com.au wrote:
Hi Brad
I agree with what Mark and Zachary have said and will expand on these.
Firstly, shard and index level operations in ElasticSearch are
peer-to-peer. Single-shard
Don't try ec2 discovery until you have tested that:
- you can connect from one machine to another on port 9300 ( nc as client
and server, basic networking/ firewalling)
- run a simple aws ec2 describe instances call with the API key you plan to
use, and you can see the machines you need there.
Agreed is bad form to force reinstall.but surely you would have your
yml in a code/cfg repository?
On 18/02/2014 9:14 AM, Tony Su tonysu...@gmail.com wrote:
What?!
Removing and re-installing the ES package either removes the original or
over-writes the existing elasticsearch.yml
The is
As I understand it, the ec2 plugin simply does ec2 API calls to list
instances, filtered as per your config. It plays no actual part in the
connectivity or clustering part - just discovery.
So yes it makes sense what you saw.
On 23/01/2014 7:43 PM, barak barak.ya...@gmail.com wrote:
Hi,
I've
Brad,
What version of ES?
What was ES doing on the problem node? (Hot threads call/ log file/
strace). Any related OS info( was it io bound? )
If it was really hung, I am not sure why the shutdown would work after
moving the shards off it ( I.e. cluster was green...) ...it sounds to me
like it was
Gotcha, my bad.
On 17/01/2014 12:43 AM, joergpra...@gmail.com joergpra...@gmail.com
wrote:
minimum_master_nodes is a dynamic cluster setting, that means, it can be
set via cluster update API.
Jörg
--
You received this message because you are subscribed to the Google Groups
elasticsearch
+1 having a queue and consumers between your source of truth and ES is a
great approach. You cab decouple and independently scale ( and stop when
needed as DP said) the different components, minimising impact to your
users.
On 09/01/2014 7:35 AM, David Pilato da...@pilato.fr wrote:
I would do 1/
29 matches
Mail list logo