lucene.store.type=ram

[9年前 に Brian Gormanly によって更新されました。]

I have been working on optimizing lucene for a cluster. Following the documentation I will be synchronizing the indexes on a shared file location, for the increased performance over using jdbc. But in going through the portal.properties i noticed there is a lucene.store.type=ram, I do not think it would be a good configuration for a cluster as each would have it's own in-memory index, but I did not see many hits on google or in the forums on the ram setting. Anyone using or have used this configuration in the past?

[9年前 に David H Nebinger によって更新されました。]

nah, I'd avoid this myself. Liferay puts a lot of stuff into Lucene and storing the whole index in ram is going to chew up a bucket of memory. Would be the best performing option if you had millions of users and unlimited memory, though.