Hello Alexandre,
I tried to reproduce this bug, and I believe it has been fixed.
I started a i3.4xlarge instance on AWS, with Xenial as the distro:
$ uname -rv
4.4.0-1112-aws #124-Ubuntu SMP Fri Jul 24 11:10:25 UTC 2020
>From there, I checked the NVMe disks:
$ lsblk
NAMEMAJ:MIN RM SIZE RO
Update to my previous comment ... running fstrim on 4.15.0-58-generic
result in much more reported trimmed data (206.6 GiB) vs
4.15.0-55-generic (144.4 GiB) (e.g. this was not caused by snapshot
release). On 4.15.0-58-generic I see errors like this in kernel log:
BTRFS warning (device dm-0): faile
I can confirm this bug on Ubuntu 16.04.6 LTS. Apparently it is related
to latest kernel update (in my case HWE).
---
Affected version:
Linux version 4.15.0-58-generic (buildd@lgw01-amd64-037) (gcc version
5.4.0 20160609 (
I did some quick testing with some different kernel versions on Ubuntu
14.04, the issue seems to exist in versions up to 4.10 but with 4.11 it
is fixed.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/17
Status changed to 'Confirmed' because the bug affects multiple users.
** Changed in: linux-aws (Ubuntu)
Status: New => Confirmed
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1756315
Title:
I ran into this issue on 14.04 and 16.04 as well. Ubuntu 18.04 does not
seem to be affected.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1756315
Title:
fstrim and discard operations take too long
Hello David!
I launched an instance and tried to create a RAID0 array using the chunk
size of 4096 bytes. The result was the same, mkfs.xfs took around 2
hours and the same for fstrim -v on the mount point where /dev/md0 was
mounted.
--
You received this bug notification because you are a member
Can you try the following?
mdadm --create --verbose /dev/md0 -l0 -n2 --chunk 4096 /dev/nvme0n1
/dev/nvme1n1
The modern SSD (As available in the I2/I3 instance types) and the HDD
(As found in D2) have a native sector size of 4096 Bytes, and that
matches sector size for Linux LVM and pages.
I thin
