I have an environment set up running ElasticSearch 0.90.2 with two master
nodes (each on their own server) and 8 data nodes (two nodes on 4 servers):
master 1 - 4 vcpus, 32 GB RAM
master 2 - 8 vcpus, 16 GB RAM
data 1 - 4 each have 4 vcpus and 32 GB RAM
Initially, I was using a BulkRequestBuilder api to load json files to ES
but was having problems with the process hanging. So to isolate the
problem, I found the largest file (438MB) and attempted to upload via the
command line using curl. This also hung, so I tried the next largest file
that I had which was 183MB and curl was able to post this file without any
problems and only took about 12 minutes to complete.
The 453MB json file that I'm having trouble with contains binary data (the
183MB file also had binary data) and the curl command I am using:
I have an environment set up running Elasticsearch 0.90.2 with two master
nodes (each on their own server) and 8 data nodes (two nodes on 4 servers):
master 1 - 4 vcpus, 32 GB RAM
master 2 - 8 vcpus, 16 GB RAM
data 1 - 4 each have 4 vcpus and 32 GB RAM
Initially, I was using a BulkRequestBuilder api to load json files to ES
but was having problems with the process hanging. So to isolate the
problem, I found the largest file (438MB) and attempted to upload via the
command line using curl. This also hung, so I tried the next largest file
that I had which was 183MB and curl was able to post this file without any
problems and only took about 12 minutes to complete.
The 453MB json file that I'm having trouble with contains binary data (the
183MB file also had binary data) and the curl command I am using:
I did increase the http.max_content_length when I first tried to curl this
file because I got an error in the logs. I've also tried using the
--data-binary switch vice -d, but had the same results.
I have an environment set up running Elasticsearch 0.90.2 with two master
nodes (each on their own server) and 8 data nodes (two nodes on 4 servers):
master 1 - 4 vcpus, 32 GB RAM
master 2 - 8 vcpus, 16 GB RAM
data 1 - 4 each have 4 vcpus and 32 GB RAM
Initially, I was using a BulkRequestBuilder api to load json files to ES
but was having problems with the process hanging. So to isolate the
problem, I found the largest file (438MB) and attempted to upload via the
command line using curl. This also hung, so I tried the next largest file
that I had which was 183MB and curl was able to post this file without any
problems and only took about 12 minutes to complete.
The 453MB json file that I'm having trouble with contains binary data
(the 183MB file also had binary data) and the curl command I am using:
and after more than an hour, the progress meter shows:
% Total % Received % Xferd Average Speed Time
Curr.
Dload Upload Total Current Left
Speed
0 437M 0 0 0 0 0 0 --:--:-- --:--:-- --:--:--
0< HTTP/1.1 100 Continue } [data not shown]
100 437M 0 0 100 437M 0 83466 1:31:41 1:31:41 --:--:--
0
I've let this process run for more than 5 hours but it never completes
and never errors. Any ideas why ES is hanging?
--
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.
It looks you had tweaked the default settings, but I would upload data
portions of 100M at most, the whole length of the upload will require
valuable space on the heap.
I upgraded ES to 0.90.7 and it no longer hangs which is great but I get an
out of memory error after about 7-8 minutes of processing even though i've
allocated 14g max memory to all data nodes.
It looks you had tweaked the default settings, but I would upload data
portions of 100M at most, the whole length of the upload will require
valuable space on the heap.
do you have a stacktrace to look at? Also, your master nodes usually dont
need so much HEAP (it can be even counter productive, as this may lead to
longer garbage collections due to so much memory needing to be cleaned up).
I upgraded ES to 0.90.7 and it no longer hangs which is great but I get an
out of memory error after about 7-8 minutes of processing even though i've
allocated 14g max memory to all data nodes.
It looks you had tweaked the default settings, but I would upload data
portions of 100M at most, the whole length of the upload will require
valuable space on the heap.
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.