Slow performance when switching from elasticsearch c 1.4 -> 2.0


(Denis A. Egorov) #1

Hello.

I have a cluster of three machines, until yesterday worked on ES 1.4 and the last upgrade to 2.0 and now in shock ...

I do not know what to do.

What was once performed for 5 minutes now occupies 30-40 minutes ...

Where "dig"?

The last execution:
69,33s user c7,00s system 2% cpu 46: 44,24 total
That maxim that yesterday performed for 5 minutes.

Query will not change.

The sample is simple and update fields.

Thank you.

P.S.
Sorry for my English.


(Adrien Grand) #2

Can you provide us with some simple queries that now perform worse than in 1.4?


(Denis A. Egorov) #3

This is a normal update by _id .


(Adrien Grand) #4

Oh, I thought this was a search workload, but this is actually about updating. Can you capture hot threads while your test is running? https://www.elastic.co/guide/en/elasticsearch/reference/current/cluster-nodes-hot-threads.html


(Michael McCandless) #5

Are you using the bulk indexing API? How many documents in each bulk request?


(Denis A. Egorov) #6

wow! )

Thank you, bulk-API "returned the speed!"

2 min!!!

Thank you so much!
Prosperity to you.


(system) #7