Luke with Lucene 3.5.0


(phobos182) #1

Trying to use Luke 3.4.0 to debug some Lucene indexes, and having no luck. Getting this error.

"No valid directory at this location, try another location."

Using this command.

java -cp lukeall-3.4.0_1.jar org.getopt.luke.Luke -index /tmp/2/media_2011_44/2/index/

Any advice?


(phobos182) #2

Forgot to mention i'm using ElasticSearch 18.4


(Goog Jobs) #3

yes , i tried and failed.

在 2011年12月9日星期五,phobos182 写道:

Trying to use Luke 3.4.0 to debug some Lucene indexes, and having no luck.
Getting this error.

"No valid directory at this location, try another location."

Using this command.

java -cp lukeall-3.4.0_1.jar org.getopt.luke.Luke -index
/tmp/2/media_2011_44/2/index/

Any advice?

--
View this message in context:
http://elasticsearch-users.115913.n3.nabble.com/Luke-with-Lucene-3-5-0-tp3571556p3571556.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(Karussell) #4

This should work for ES 0.18.4 (I tried yesterday :wink: - I tried via UI
and open dir)

Be sure you are opening one index and the correct folder:

es-installation/data/yourcluster/nodes/0/indices/yourindex

Regards,
Peter.

On 9 Dez., 10:40, goog cheng googch...@gmail.com wrote:

yes , i tried and failed.

在 2011年12月9日星期五,phobos182 写道:

Trying to use Luke 3.4.0 to debug some Lucene indexes, and having no luck.
Getting this error.

"No valid directory at this location, try another location."

Using this command.

java -cp lukeall-3.4.0_1.jar org.getopt.luke.Luke -index
/tmp/2/media_2011_44/2/index/

Any advice?

--
View this message in context:
http://elasticsearch-users.115913.n3.nabble.com/Luke-with-Lucene-3-5-...
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(Karussell) #5

BTW: what has this todo with lucene 3.5 ?


(Karussell) #6

Hmmh, I originally just wanted to confirm it but it really does not
work. The version where it worked was 0.18.3. Strange, I'll check

Peter.

On 9 Dez., 10:44, Karussell tableyourt...@googlemail.com wrote:

BTW: what has this todo with lucene 3.5 ?


(Karussell) #7

Uhm, stupid me - the folder is:

es-installation/data/yourcluster/nodes/0/indices/yourindex/0

not

es-installation/data/yourcluster/nodes/0/indices/yourindex/

Peter.

On 9 Dez., 10:51, Karussell tableyourt...@googlemail.com wrote:

Hmmh, I originally just wanted to confirm it but it really does not
work. The version where it worked was 0.18.3. Strange, I'll check

Peter.

On 9 Dez., 10:44, Karussell tableyourt...@googlemail.com wrote:

BTW: what has this todo with lucene 3.5 ?


(Shay Banon) #8

Note, Lucene 3.5.0 is part of 0.18.5, 0.18.4 works with lucene 3.4.0. Luke
works well.

On Fri, Dec 9, 2011 at 11:57 AM, Karussell tableyourtime@googlemail.comwrote:

Uhm, stupid me - the folder is:

es-installation/data/yourcluster/nodes/0/indices/yourindex/0

not

es-installation/data/yourcluster/nodes/0/indices/yourindex/

Peter.

On 9 Dez., 10:51, Karussell tableyourt...@googlemail.com wrote:

Hmmh, I originally just wanted to confirm it but it really does not
work. The version where it worked was 0.18.3. Strange, I'll check

Peter.

On 9 Dez., 10:44, Karussell tableyourt...@googlemail.com wrote:

BTW: what has this todo with lucene 3.5 ?


(phobos182) #9

I just tried again, and it's not working for me. I am opening the correct folder. I am using data.path, and have my data in ES striped across multiple hard drives on a node. I'm wondering if that has anything to do with the problem.


(phobos182) #10

Receiving this error.

java.io.FileNotFoundException: /data/0/elasticsearch/es_cluster1/nodes/0/indices/media_2011_36/14/index/_wue.fnm (No such file or directory)


(Shay Banon) #11

Yes, if you stripe the data over several directories, then luke will not
work since the different index files are placed on different directories.

On Sat, Dec 10, 2011 at 1:33 AM, phobos182 phobos182@gmail.com wrote:

Receiving this error.

java.io.FileNotFoundException:

/data/0/elasticsearch/es_cluster1/nodes/0/indices/media_2011_36/14/index/_wue.fnm
(No such file or directory)

--
View this message in context:
http://elasticsearch-users.115913.n3.nabble.com/Luke-with-Lucene-3-5-0-tp3571556p3574418.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(phobos182) #12

I have found a solution for this. Basically look for all instances of a index in the data.path, and copy them to a working directory. Then you have it all basically merged down to one copy.

Example.

find %{data.path} -type d -name '%{index}' | grep %{shard} | xargs -I{} cp -rp {} /tmp/2/

Working
find . -type d -name '2011-11-14' | grep 14 | xargs -I{} cp -rp {} /tmp/2/

Then use luke on /tmp/2/14


(Shay Banon) #13

Copying over several shards to teh same location will override each other
and might create a corrupted index, just look at the single level shard
index.

On Mon, Dec 12, 2011 at 6:03 PM, phobos182 phobos182@gmail.com wrote:

I have found a solution for this. Basically look for all instances of a
index
in the data.path, and copy them to a working directory. Then you have it
all
basically merged down to one copy.

Example.

find %{data.path} -type d -name '%{index}' | grep %{shard} | xargs -I{} cp
-rp {} /tmp/2/

Working
find . -type d -name '2011-11-14' | grep 14 | xargs -I{} cp -rp {} /tmp/2/

Then use luke on /tmp/2/14

--
View this message in context:
http://elasticsearch-users.115913.n3.nabble.com/Luke-with-Lucene-3-5-0-tp3571556p3579833.html
Sent from the ElasticSearch Users mailing list archive at Nabble.com.


(system) #14