Split indexing and scheduling into two separate processes under the queue server
administrator2011-12-07 11:13
Right now they execute serially in a loop and so scheduling can actually slow down/block the indexing process. Running in separate processes would allow for better parallelism and hopefully more pages indexed/hour.
The patch should be backward compatible and work if someone hasn't done the necessary configurations to make the manage machine activity work.
administrator2012-01-31 00:43
This was implemented over several commits, but the first pass was largely completed by 340f9bab
Right now they execute serially in a loop and so scheduling can actually slow down/block the indexing process. Running in separate processes would allow for better parallelism and hopefully more pages indexed/hour.
The patch should be backward compatible and work if someone hasn't done the necessary configurations to make the manage machine activity work.
This was implemented over several commits, but the first pass was largely completed by 340f9bab