Something is wrong with my elasticsearch instance. It complains about
unassigned shards en cluster health is red.
I was wondering if it is possible to do complete removal of Elasticsearch.
That way I can easily recover from the problems by removal and install.
This however does seem to be straightforward to the point that it makes me
wonder if it is at all possible.
This is under Ubuntu 14.04 something and elasticsearch 1.4.2.
Using Ubuntu software centre I can remove 1.4.2 but directory remains. So
normal software remove does not seem to do a lot.
If I combine that with removal of directories left by Elasticsearch
scattered throughtout the file system, it seems to end with a corruption of
some sort. Elasticsearch won't install.
Setting up elasticsearch (1.4.2) ...
chown: cannot access /etc/elasticsearch/*: No such file or directory
So the installation program expect a directory to exist prior to installation which is a bit strange. I would argue that one should either cleanup file system on uninstall 'or' don't make assumption on what it means if directories are present. It should be one of two not a bit of both.
Is there a documented way to recover from this that does not involve me doing a complete reinstall of the OS?
If you want complete reinstall of Elasticsearch you need to reinstall the
OS. It is the only way. Elasticsearch was designed in such way that it
requires full OS reinstall. If there is a corruption, this is what you
should do. It is straightforward enough.
On Tuesday, 10 February 2015 16:59:34 UTC+1, Onno wrote:
Something is wrong with my elasticsearch instance. It complains about
unassigned shards en cluster health is red.
I was wondering if it is possible to do complete removal of Elasticsearch.
That way I can easily recover from the problems by removal and install.
This however does seem to be straightforward to the point that it makes me
wonder if it is at all possible.
This is under Ubuntu 14.04 something and elasticsearch 1.4.2.
Using Ubuntu software centre I can remove 1.4.2 but directory remains. So
normal software remove does not seem to do a lot.
If I combine that with removal of directories left by Elasticsearch
scattered throughtout the file system, it seems to end with a corruption of
some sort. Elasticsearch won't install.
Setting up elasticsearch (1.4.2) ...
chown: cannot access /etc/elasticsearch/*: No such file or directory
So the installation program expect a directory to exist prior to installation which is a bit strange. I would argue that one should either cleanup file system on uninstall 'or' don't make assumption on what it means if directories are present. It should be one of two not a bit of both.
Is there a documented way to recover from this that does not involve me doing a complete reinstall of the OS?
No, it does not require you to reinstall the OS.
You can use apt-get purge elasticsearch to remove the package and any
directories and files it created.
Also, reinstalling the OS just to fix your problem is a major waste of
time, you should have tried to restore your cluster to green.
If you want complete reinstall of Elasticsearch you need to reinstall the
OS. It is the only way. Elasticsearch was designed in such way that it
requires full OS reinstall. If there is a corruption, this is what you
should do. It is straightforward enough.
On Tuesday, 10 February 2015 16:59:34 UTC+1, Onno wrote:
Something is wrong with my elasticsearch instance. It complains about
unassigned shards en cluster health is red.
I was wondering if it is possible to do complete removal of
Elasticsearch. That way I can easily recover from the problems by removal
and install. This however does seem to be straightforward to the point that
it makes me wonder if it is at all possible.
This is under Ubuntu 14.04 something and elasticsearch 1.4.2.
Using Ubuntu software centre I can remove 1.4.2 but directory remains. So
normal software remove does not seem to do a lot.
If I combine that with removal of directories left by Elasticsearch
scattered throughtout the file system, it seems to end with a corruption of
some sort. Elasticsearch won't install.
Setting up elasticsearch (1.4.2) ...
chown: cannot access /etc/elasticsearch/*: No such file or directory
So the installation program expect a directory to exist prior to installation which is a bit strange. I would argue that one should either cleanup file system on uninstall 'or' don't make assumption on what it means if directories are present. It should be one of two not a bit of both.
Is there a documented way to recover from this that does not involve me doing a complete reinstall of the OS?
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.