Thanks for the reply Boaz. Sorry for the late reply, was away for a while.
We run ES ver 1.2.2 and use the zulu openJDK from azul.
So I'm going to get you the mappings in another post, but I think the
cluster ended up becoming corrupt from an incident we saw about 1 1/2 weeks
ago. I started another discussion about it
I'm thinking the continuous updating of mappings (which caused the internal
DOS on the cluster) possibly had a sequence where it lead to a bad update
of the state file.
The state file for that incident was pretty big (70MB+), I'm not sure if
you want that. I can probably get a scoped down version of the mappings we
had after the incident.
On Saturday, August 30, 2014 11:23:21 AM UTC-7, Boaz Leskes wrote:
Creating a broken mapping shouldn't be possible and sounds like a serious
bug. Can you gist your broken mapping ? I would like to see if I can figure
out how it happened (if you know it will be great). Also - what version of
ES are you using?
I hope that seeing the mapping I can up with a work around. There are a
couple of other options but choosing one depends on the amount data you
have and wether you're OK with some down time.
On Thursday, August 28, 2014 11:55:04 PM UTC+2, Joshua Montgomery wrote:
Is it possible to fix the state file if its been corrupted. Somehow a bad
mapping was added and it had an extra '}' throwing the entire state file
off. Deleting the bad mapping didn't fix the issue. Do I need to rebuild
the entire cluster or is there a quick fix? Should Elasticsearch even allow
Also is it a good idea to run the cluster with this issue.
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 email@example.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/03ee610a-bc90-405d-8153-736e7e793f7b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.