Hello all,
I have a huge elasticsearch cluster with around 150 Tb of data indexed into
it. It has 32 nodes and more than 3000 shards in active state. I am
planning to upgrade this elasticsearch cluster from version 0.90.5 to
1.0.1. Are there any index compatibility issues that I need to be aware
of? I know that 0.90.5 is based on lucene 4.4 and 1.0.1 is based on lucene
4.6.1 (correct me if im wrong) I am not quite sure about the codecs
compatibility between these versions of lucene indices. Can anybody shade a
light on it?
Are there any pointers or issues that I need to look out for in order to do
the upgrade? Also I am not planning to do a rolling upgrade. So I am not
worried about transport client compatibility issues.
Thanks in advance!!
On 27 February 2014 20:00, hrishikesh prabhune hruship@gmail.com wrote:
Hello all,
I have a huge elasticsearch cluster with around 150 Tb of data indexed
into it. It has 32 nodes and more than 3000 shards in active state. I am
planning to upgrade this elasticsearch cluster from version 0.90.5 to
1.0.1. Are there any index compatibility issues that I need to be aware
of? I know that 0.90.5 is based on lucene 4.4 and 1.0.1 is based on lucene
4.6.1 (correct me if im wrong) I am not quite sure about the codecs
compatibility between these versions of lucene indices. Can anybody shade a
light on it?
Are there any pointers or issues that I need to look out for in order to
do the upgrade? Also I am not planning to do a rolling upgrade. So I am not
worried about transport client compatibility issues.
Thanks in advance!!
You should take note of the compatibility breaking changes in the release
notes of course and do some functional tests with the new version to ensure
that you are not affected by those changes. Additionally, I would carefully
plan and test the update procedure with a smaller cluster first. In theory,
you should be able to do a rolling restart with the new version but that
still is a somewhat risky procedure where there is some room for unexpected
outages. Doing upgrades like this without any downtime can be challenging.
The easiest is to just shutdown the whole cluster, upgrade all the nodes
and then bring it back up. When I did this a few weeks ago with my logstash
cluster (0.90.8 -> 1.0.0), it spent about 15 minutes converting my indices
before the cluster went green again. That was a small setup with only 10GB.
The differences between 0.90.5 and 1.0.0 are not such that I would perform
any upgrades to in between versions. The risks for those upgrades are
similar to doing the proper upgrade to 1.0.0 directly probably. You'd have
to deal with several lucene updates and index conversions.
Jilles
On Thursday, February 27, 2014 10:00:45 AM UTC+1, hrishikesh prabhune wrote:
Hello all,
I have a huge elasticsearch cluster with around 150 Tb of data indexed
into it. It has 32 nodes and more than 3000 shards in active state. I am
planning to upgrade this elasticsearch cluster from version 0.90.5 to
1.0.1. Are there any index compatibility issues that I need to be aware
of? I know that 0.90.5 is based on lucene 4.4 and 1.0.1 is based on lucene
4.6.1 (correct me if im wrong) I am not quite sure about the codecs
compatibility between these versions of lucene indices. Can anybody shade a
light on it?
Are there any pointers or issues that I need to look out for in order to
do the upgrade? Also I am not planning to do a rolling upgrade. So I am not
worried about transport client compatibility issues.
Thanks in advance!!
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.