Things you can do are trying to track down if there is any pattern -
are these first request on a child, "nth" request on child etc...
we add additional variables to access logs which include things such
as PID/request no in PID, memory usage before and after etc. This
allows us sometimes to see patterns in errors which aren't normally
easy to see.. Saying that it may not help you here... Another thing you can do with ab is check to see if it is OK under lower load? Finally I find "siege" better than "ab" when it comes to testing as it is more configurable - you can send it a list of URLs - run test for "n-seconds" rather than "n-requests" which is more useful if you don't know how long things take to return! James On 02/10/2016 09:02, Daniel wrote:
-- The Wellcome Trust Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE. |