Re: Problems maintaining large high-availability search

2004-06-08 Thread Atle Veka
Doesn't look like my reply last night made it to the list, resend... Hello Seth, On Mon, 7 Jun 2004, Seth Brundle wrote: 1. The timeout is set to 5 min, because of the number of queries, there are a lot of unused http processes that linger with connections, and the only way to seeminly

Problems maintaining large high-availability search

2004-06-07 Thread Seth Brundle
I have been maintaining a mod-perl/MySQL4 web search engine with about 3M records, which gains and drops about 100k records daily. It runs on a dual-Opteron 242 system with 8GB RAM, 15k SCSI driv, SUSE Professional for AMD64. It recently grew to this size (from about 1M records), and I am

Re: Problems maintaining large high-availability search

2004-06-07 Thread Fagyal Csongor
Hi, 1. The timeout is set to 5 min, because of the number of queries, there are a lot of unused http processes that linger with connections, and the only way to seeminly keep MySQL connections available is to keep timeouts short. What about using a connection-pool? Like Apache::DBI. It should

RE: Problems maintaining large high-availability search

2004-06-07 Thread Donny Simonton
1. The timeout is set to 5 min, because of the number of queries, there are a lot of unused http processes that linger with connections, and the only way to seeminly keep MySQL connections available is to keep timeouts short. What about using a connection-pool? Like Apache::DBI. It