On 5/9/07, Quanah Gibson-Mount quanah@zimbra.com wrote:
--On May 9, 2007 2:51:23 PM -0400 Sam Tran stlist@gmail.com wrote:
On 5/9/07, Quanah Gibson-Mount quanah@zimbra.com wrote:
--On May 9, 2007 12:46:18 PM -0400 Sam Tran stlist@gmail.com wrote:
On 5/8/07, Quanah Gibson-Mount quanah@zimbra.com wrote: [snip]
How have you tuned DB_CONFIG for the accesslog backend?
--Quanah
Here is my DB_CONFIG for the accesslog backend:
set_cachesize 0 52428800 0 set_lg_regionmax 1048576 set_lg_max 10485760 set_lg_bsize 2097152 set_lg_dir /var/log/openldap_bdb/accesslog set_tmp_dir /tmp
Have you looked at the output of db_stat -m to see how things are behaving?
I ran 1,000 sequential writes on the provider. Then run the command you specified. Here is the output:
According to the output most of the data is in cache. It took about 5 minutes for the consumer to get the updates. I still don't know why.
Well, things seem fine from the accesslog DB side. On my servers @ Stanford, the largest latency I've seen is about 7 seconds. I wonder if something is slow on your network.
I don't think the network speed is the problem here. I am running a host-based firewall on the consumer server. I only allow traffic on port 636 between the consumer and the provider. Do I need to open other holes for delta-syncrepl to work?
Thanks, Sam