JRE Crashing while indexing

Hey guys,

I've been trying out ES for the past few days and keep running into an
issue with the JRE crashing on me while indexing. This first happened
while running 1.6.0_26, so I updated to the most recent version,
1.7.0_01, which also crashed on me.

I'm indexing 4 million docs [which are ~600 bytes each], in batches of
50,000 using the _bulk endpoint. I'm indexing using the Elastica
library for PHP. Towards the end of the indexing process, it crashes
[and PH emits a few generic exceptions from Elastica].

I checked the log files and found some stuff in the "elasticsearch-
stdout" file, which I have linked below.

Any suggestions?

Also, I'm very new to Java so if any terminology I used is strange I
can clarify :slight_smile:

Thanks so much!


Environment:
java version "1.7.0_01"
Java(TM) SE Runtime Environment (build 1.7.0_01-b08)
Java HotSpot(TM) 64-Bit Server VM (build 21.1-b02, mixed mode)

java version "1.6.0_26"
Java(TM) SE Runtime Environment (build 1.6.0_26-b03)
Java HotSpot(TM) 64-Bit Server VM (build 20.1-b02, mixed mode)

ES 0.17.9 running as a service through Commons Daemon Package version
1.0.5.

JRE 1.6 crash file: https://gist.github.com/1302913
JRE 1.7 crash file: https://gist.github.com/1302916

Cluster info: https://gist.github.com/1302899
Index info: https://gist.github.com/1302905

Hey,

Can you only test with 1.6, as there are known problem with 1.7 (though
upadte 1 might fix the). Also, can you run without compressed opts? And if
it still fails, can you configure the index store to use simplefs (since its
windows) and not mmapfs (which is the default on windows). Just set
index.store.type to simplefs and see if it crashes.

On Fri, Oct 21, 2011 at 3:59 AM, Chris Williams chris@colourlovers.comwrote:

Hey guys,

I've been trying out ES for the past few days and keep running into an
issue with the JRE crashing on me while indexing. This first happened
while running 1.6.0_26, so I updated to the most recent version,
1.7.0_01, which also crashed on me.

I'm indexing 4 million docs [which are ~600 bytes each], in batches of
50,000 using the _bulk endpoint. I'm indexing using the Elastica
library for PHP. Towards the end of the indexing process, it crashes
[and PH emits a few generic exceptions from Elastica].

I checked the log files and found some stuff in the "elasticsearch-
stdout" file, which I have linked below.

Any suggestions?

Also, I'm very new to Java so if any terminology I used is strange I
can clarify :slight_smile:

Thanks so much!


Environment:
java version "1.7.0_01"
Java(TM) SE Runtime Environment (build 1.7.0_01-b08)
Java HotSpot(TM) 64-Bit Server VM (build 21.1-b02, mixed mode)

java version "1.6.0_26"
Java(TM) SE Runtime Environment (build 1.6.0_26-b03)
Java HotSpot(TM) 64-Bit Server VM (build 20.1-b02, mixed mode)

ES 0.17.9 running as a service through Commons Daemon Package version
1.0.5.

JRE 1.6 crash file: JRE 1.6 crash · GitHub
JRE 1.7 crash file: JRE 1.7 crash · GitHub

Cluster info: _cluster/nodes · GitHub
Index info: 1302905’s gists · GitHub

After we get installed and running for our dev purposes, I plan to see we
can be stable at the current 6 update. {This was the most relevant thread
we could find for java versions}

We certainly should be able to track the Lucene currency I would think.
Apache does a very good job of calling delays on updates that are found to
be package-affecting. Here is update 29's reference to a called delay by
Apache (and they applied it to 7.0.0 also, since 7.0.3 is the current, it
may be fixed on that line also.

Any suggestions on how to pre-certify later java versions than mentioned
here would be appreciated.

java 6 rel notes u 29http://www.oracle.com/technetwork/java/javase/6u29-relnotes-507960.html

JIT and Loop Bugs

Three bugs reported by various parties, including Apache Lucene
developers, have been fixed in JDK 6 Update 29, in addition to a fourth
related bug found by Oracle (7070134, 7068051, 7044738, 7077439).

On Thursday, October 20, 2011 10:25:08 PM UTC-4, kimchy wrote:

Hey,

Can you only test with 1.6, as there are known problem with 1.7 (though
upadte 1 might fix the). Also, can you run without compressed opts? And if
it still fails, can you configure the index store to use simplefs (since
its windows) and not mmapfs (which is the default on windows). Just set
index.store.type to simplefs and see if it crashes.

On Fri, Oct 21, 2011 at 3:59 AM, Chris Williams <ch...@colourlovers.com<javascript:>

wrote:

Hey guys,

I've been trying out ES for the past few days and keep running into an
issue with the JRE crashing on me while indexing. This first happened
while running 1.6.0_26, so I updated to the most recent version,
1.7.0_01, which also crashed on me.

I'm indexing 4 million docs [which are ~600 bytes each], in batches of
50,000 using the _bulk endpoint. I'm indexing using the Elastica
library for PHP. Towards the end of the indexing process, it crashes
[and PH emits a few generic exceptions from Elastica].

I checked the log files and found some stuff in the "elasticsearch-
stdout" file, which I have linked below.

Any suggestions?

Also, I'm very new to Java so if any terminology I used is strange I
can clarify :slight_smile:

Thanks so much!


Environment:
java version "1.7.0_01"
Java(TM) SE Runtime Environment (build 1.7.0_01-b08)
Java HotSpot(TM) 64-Bit Server VM (build 21.1-b02, mixed mode)

java version "1.6.0_26"
Java(TM) SE Runtime Environment (build 1.6.0_26-b03)
Java HotSpot(TM) 64-Bit Server VM (build 20.1-b02, mixed mode)

ES 0.17.9 running as a service through Commons Daemon Package version
1.0.5.

JRE 1.6 crash file: JRE 1.6 crash · GitHub
JRE 1.7 crash file: JRE 1.7 crash · GitHub

Cluster info: _cluster/nodes · GitHub
Index info: 1302905’s gists · GitHub

--