On Wed, 8 Jun 2022 15:52:14 +0700 Pavel <pavel2000@xxxxxxxxxxxxx> wrote: > 08.06.2022 15:32, Mariusz Tkaczyk пишет: > > Hi Pavel, > > IMO it is not a RAID problem. In this case some part of requests hangs > > inside nvme driver and raid1d hanged too. It is rather nvme problem not a > > raid. RAID should handle it well if IO errors are continuously reported. > > > > Thanks, > > Mariusz > > Hi, > > When we get I/O error on "generic" drive (not NVMe), we get device in > failed state and degraded array. > For reasons unknown to me, RAID on NVMe behaves in different way. Because some IO requests are blocked inside nvme driver and RAID engine hangs when it tries to flush next requests. In "It is rather nvme problem" I meant nvme driver in kernel. Sorry for confusion. Thanks, Mariusz
