Elasticsearch 0.90 Beta 1

Deebs

Well-known member
I converted my forum over from ES 0.20.x and OpenJava (or whatever it is called) to ES 0.90 Beta 1 and Oracle Java 1.7 update 17 without any problems last night.

As compression is on by default in ES 0.90 the forum index went from 4gb down to 1.9gb :) Anyway, just thought I would let you know it works fine.
 
I converted my forum over from ES 0.20.x and OpenJava (or whatever it is called) to ES 0.90 Beta 1 and Oracle Java 1.7 update 17 without any problems last night.

As compression is on by default in ES 0.90 the forum index went from 4gb down to 1.9gb :) Anyway, just thought I would let you know it works fine.

Good to know, will upgrade P8nt shortly and report back the results.
 
So apparently 0.90.x still has the same issue for me that 0.20.x has (can't show content by a user). Maybe has something to do with me not storing _source? Either way, from the shell, this (which is what XenForo is doing), gives 0 results:

Code:
curl -XPOST localhost:9200/xenforo/_search -d '{"query":{"filtered":{"filter":{"and":[{"term":{"user":1}},{"type":{"value":"thread"}}]},"query":{"match_all":{}}}},"sort":[{"date":"desc"}],"size":1000,"fields":["date"]}'

But making it so you are searching just the one content type (even leaving all the filtering parameters in place), it works...
Code:
curl -XPOST localhost:9200/xenforo/thread/_search -d '{"query":{"filtered":{"filter":{"and":[{"term":{"user":1}},{"type":{"value":"thread"}}]},"query":{"match_all":{}}}},"sort":[{"date":"desc"}],"size":1000,"fields":["date"]}'

I hacked on XenES_Api::search() a little bit to make it so a search with type filtering just searches the one content type index, and everything works as expected. It's pretty strange... really looks like an ES bug to me that started with 0.20.x and maybe just happens when you don't save _source?
 
Interesting side note... I just realized today that I forgot to allocate more than the default 1024MB to the ES_HEAP_SIZE when I upgraded (really annoying ES doesn't store it's config stuff in something like /etc btw...

Anyway... it's been humming along fine with the nodes just using the default 1024MB for more than a week and noticed no ill effects or anything being slower than when I allocated way more memory. Maybe the operating system is giving it "real" memory instead of disk swap or something... but thought it was worth noting.

I'm just gonna leave it at 1024... {shrug}
 
Interesting side note... I just realized today that I forgot to allocate more than the default 1024MB to the ES_HEAP_SIZE when I upgraded (really annoying ES doesn't store it's config stuff in something like /etc btw...

Anyway... it's been humming along fine with the nodes just using the default 1024MB for more than a week and noticed no ill effects or anything being slower than when I allocated way more memory. Maybe the operating system is giving it "real" memory instead of disk swap or something... but thought it was worth noting.

I'm just gonna leave it at 1024... {shrug}
Yeh that is annoying that the config file is stored in the data area and that subsequent versions can overwrite your changes. Interesting about the memory usage. Will have to check what I have mine set to when I finally get into work. (Awfully slow net connection where I am atm)
 
Back
Top Bottom