https://bugzilla.kernel.org/show_bug.cgi?id=217965 --- Comment #17 from Ojaswin Mujoo (ojaswin.mujoo@xxxxxxx) --- Hey Ivan, Eduard, Thanks for all the data. The iostat suggests that we are breaking up the writes much more as seen in the wareq-sz field, which is much lower in the regressed kernel. Further, since the issue is only when striping is used (ie in raid scenarios) it seems like it has something to do with raiding. Eduard's replicator seems much easier to try so let me see if I'm able to replicate this over a raid setup of HDDs. Regards, ojaswin -- You may reply to this email to add a comment. You are receiving this mail because: You are watching the assignee of the bug.
