[Bug 1030310] Regression in WAL and COMMIT times for pgioperf-bench on ext3

http://bugzilla.suse.com/show_bug.cgi?id=1030310 http://bugzilla.suse.com/show_bug.cgi?id=1030310#c20 --- Comment #20 from Jan Kara <jack@suse.com> --- Regarding the read starvation - I didn't observe anything like that in my testing although I do agree that after my patches there were less read samples since writes weren't backlogged as much behind blocked journal. I'll try to run your version of the benchmark, see what happens and possibly come up with better parameters for the benchmark. -- You are receiving this mail because: You are on the CC list for the bug.