RE: [jira] Updated: (SOLR-240) java.io.IOException: Lock obtain timed out: SimpleFSLock

2007-07-13 Thread Will Johnson
comments?

Hooray, and very cool.  I didn't know you only needed a locking
mechanism if you only have multiple index writers so the use of NoLock
by default makes perfect sense.

A quick stability update: Since I first submitted the patch ~2 months
ago we've had 0 lockups with it running in all our test environments.  

- will


Re: [jira] Updated: (SOLR-240) java.io.IOException: Lock obtain timed out: SimpleFSLock

2007-07-13 Thread Yonik Seeley

On 7/13/07, Will Johnson [EMAIL PROTECTED] wrote:

Hooray, and very cool.  I didn't know you only needed a locking
mechanism if you only have multiple index writers so the use of NoLock
by default makes perfect sense.


For Lucene, you do (did.. before lockless commits pach) need locking
(a read lock) even to open an index with a reader.  The write lock is
still also needed to avoid a reader changing the index via deletion at
the same time a writer is.  Solr coordinates this at a higher level,
hence it's not really needed.

-Yonik