On Jan 13, 2021, at 10:19 AM, Jan Kara <jack@xxxxxxx> wrote: > > Hi! > > On Tue 12-01-21 19:45:06, Xiaoguang Wang wrote: >> I use io_uring to evaluate ext4 randread performance(direct io), observed >> obvious overhead in jbd2_transaction_committed(): I was going to ask about this - is the filesystem mounted with noatime or relatime or lazytime? Otherwise, it may be a lot of atime updates that are causing all of this journal traffic in what _should_ be a read-only workload? Cheers, Andreas
Attachment:
signature.asc
Description: Message signed with OpenPGP
