Lucene 4.1 AnalyzingSuggester in ES having persistable FST

FST built for my application is large. Is there any work done by someone to store FST generated to index repository or to any other persistent cache

Is this true? FST are not persistent, they do not survive node restarts?

Jörg

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

I believe Lucene's default implementation is in-memory. Someone please
confirm whether elasticsearch has overriden this behavior

On Thursday, September 12, 2013 1:26:25 PM UTC+5:30, Jörg Prante wrote:

Is this true? FST are not persistent, they do not survive node restarts?

Jörg

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

In Elasticsearch, the analyzing suggester writes its FST per-segment at
index time. So we can just load the FST from disk the first time it is
used.

clint

On 12 September 2013 10:44, Chetana ambha.career@gmail.com wrote:

I believe Lucene's default implementation is in-memory. Someone please
confirm whether elasticsearch has overriden this behavior

On Thursday, September 12, 2013 1:26:25 PM UTC+5:30, Jörg Prante wrote:

Is this true? FST are not persistent, they do not survive node restarts?

Jörg

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

The FSTs are persistent on disk - they can be recovered via the ES recovery
mechanism and they are durable. The FSTs are in memory when you read them
at query time but that doesn't mean they are lost on node restart. The
Lucene index also uses FSTs which are written to disk in the term
dictionary and when you open the index they are pulled into memory.

simon

On Thursday, September 12, 2013 2:33:33 PM UTC+2, Clinton Gormley wrote:

In Elasticsearch, the analyzing suggester writes its FST per-segment at
index time. So we can just load the FST from disk the first time it is
used.

clint

On 12 September 2013 10:44, Chetana <ambha....@gmail.com <javascript:>>wrote:

I believe Lucene's default implementation is in-memory. Someone please
confirm whether elasticsearch has overriden this behavior

On Thursday, September 12, 2013 1:26:25 PM UTC+5:30, Jörg Prante wrote:

Is this true? FST are not persistent, they do not survive node restarts?

Jörg

--
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 elasticsearc...@googlegroups.com <javascript:>.
For more options, visit https://groups.google.com/groups/opt_out.

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.