On Mar 17, 2011, at 3:19 PM, Shawn Heisey wrote: On 3/17/2011 3:43 AM, Vadim Kisselmann wrote: Unfortunately, this doesn't seem to be the problem. The queries themselves are running fine. The problem is that the replications is crawling when there are many queries going on and that the replication speed stays low even after the load is gone.
If you run "iostat 5" what are typical values on each iteration for the various CPU states while you're doing load testing and replication at the same time? In particular, %iowait is important. CPU stats from top (iostat doesn't seem to show CPU load correctly): 90.1%us, 4.5%sy, 0.0%ni, 5.5%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st Seems like I/O is not the bottleneck here. Other interesting thing: When Solr starts its replication under heavy load, it tries to download the whole index from master. >From /solr/admin/replication/index.jsp: Current Replication Status Start Time: Thu Mar 17 15:57:20 CET 2011 Files Downloaded: 9 / 163 Downloaded: 83,04 MB / 97,75 GB [0.0%] Downloading File: _d5x.nrm, Downloaded: 86,82 KB / 86,82 KB [100.0%] Time Elapsed: 419s, Estimated Time Remaining: 504635s, Speed: 202,94 KB/s