@xnox: I have:
$ uname -r
5.15.0-43-generic
$ sudo zfs version
zfs-2.1.4-0ubuntu0.1
zfs-kmod-2.1.4-0ubuntu0.1
Can you confirm that the bug is not fixed there, since presumably kernel
5.15.0-43 < 5.15.0-43.46?
And is it possible to upgrade the kernel within jammy to get the fix?
--
You
Hi, the zfs userspace .deb package versions can be out of sync with the
kernel driver zfs kernel module version number. This can be seen in the
output of
$ zfs version
For example on my test system I have:
$ sudo zfs version
zfs-2.1.4-0ubuntu0.1
zfs-kmod-2.1.5-1ubuntu2
Because I have a newer
@lucaskanashiro: I do not have innodb_page_size set.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on
Hi Sergio
Thanks for the response.
I'm actually running a desktop installation of Jammy (Ubuntu MATE),
originally 20.04 and upgraded through the non-LTS releases.
After a dist-upgrade the other day I found that the mysql-server-8.0
package could not be configured and I found errors in the error
Hi Michael,
In order to confirm whether we're talking about the same bug here, I
tried to reproduce it. Here's what I did:
1) Grab a Focal machine, up-to-date.
2) Run "lxd init" and make sure to choose "zfs" as the backend format
for its storage.
3) Fire up a Focal VM using "lxc launch
@lucaskanashiro
> However, the default of innodb_page_size is 16kb
> (https://dev.mysql.com/doc/refman/8.0/en/server-status-variables.html#statvar_Innodb_page_size),
> and 1712128/16 = 107008. Did you change the innodb_page_size global
> value by any chance? If you can share your configuration it
Until such time the fix makes it into mainstream, this hack gets things
going again for me.
I took a look at the contents of the directory /var/lib/mysql/#innodb_redo#
There was only a single file there which of course matched the filename
identified in the /var/log/mysql/error.log and this
@mrichtarsky from the logs, mysql was not able to redo the actions
during the upgrade because the size of the ib_redo6 file was not
multiple of innodb_page_size. However, the default of innodb_page_size
is 16kb (https://dev.mysql.com/doc/refman/8.0/en/server-status-
There is no task for this bug targeting Ubuntu Focal (20.04), that's why
it is not fixed there. Apparently, no one reported this issue there,
only in Jammy (22.04). @shadowlmd could you confirm if you are facing
the very same issue described on this bug in Focal? If it is another
issue please file
I ran into this today on jammy, which I understand should have the fix?
Here are my symptoms:
- unattended-upgrades upgraded mysql server:
$ grep upgrade /var/log/dpkg.log | grep mysql
2022-07-30 06:38:00 upgrade mysql-server:all 8.0.29-0ubuntu0.22.04.2
8.0.30-0ubuntu0.22.04.1
2022-07-30
Also, MySQL 8.0.30 was released recently in Ubuntu 20.04. This version
uses falloate() for redo files. This leads to very hard to recover
corruption after MySQL server restart.
https://bugs.mysql.com/bug.php?id=108012
Please fix this in Ubuntu 20.04 ASAP.
** Bug watch added: MySQL Bug System
This bug is still not fixed in Ubuntu 20.04.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on zfs in
The fix for this bug is already verified but the same upload also
contains the fix for this bug which is still not verified:
https://bugs.launchpad.net/ubuntu/+source/zfs-linux/+bug/1969482
That's why it is blocked in jammy-proposed at the moment.
--
You received this bug notification because
In deed. It's a good question why it is taking so long between proposed
and updates repositories. Anyone knows why the delay?
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
The bug was fixed in 2.1.4-0ubuntu0.1 but that version is still in
proposed for Jammy. The current release version still seems to be
2.1.2-1ubuntu3.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
@lvoytek
The fix was released 19th, April
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on zfs in
Are there any updates on getting this released into Jammy? We've gotten
several reports of mysql failing due to this issue
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Ah alright, that's fine then. Hopefully the Jammy version can move from
proposed to release soon.
Thanks!
** Tags removed: verification-failed-jammy
** Tags added: verification-done-jammy
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to
Once an updated kernel is made available, this will just work.
On Mon, 23 May 2022, 18:40 Lena Voytek, <1969...@bugs.launchpad.net>
wrote:
> That's fair, it does seem to work when using that method. However, for
> this to be an actual fix it needs to work out of the box. Since mysql
> still
That's fair, it does seem to work when using that method. However, for
this to be an actual fix it needs to work out of the box. Since mysql
still fails to install and fallocate still fails on kinetic after
installing with zfs, it isn't ready for normal users yet
--
You received this bug
Unless you've also installed zfs-dkms and built the ZFS kernel modules
using DKMS, you've not installed the fix completely on your system.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
Added the apport output of kinetic in its current state to a duplicate
bug: https://pad.lv/1975513
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with
I've done some more testing with Jammy and can confirm this does not
fully fix the issue. I also tried to install mysql on kinetic now that
the fix is 'released' and it still fails. I think more needs to be done
with this to get fallocate working on zfs
** Tags removed: verification-done-jammy
**
It still doesn't work. I'm not able to start mysql-server or mariadb on
Ubuntu 22.04 LXC Container running on Proxmox 7.2-3 with ZFS.
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
** Tags added: verification-done-jammy
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on zfs in
Jammy
This bug was fixed in the package zfs-linux - 2.1.4-0ubuntu0.1
---
zfs-linux (2.1.4-0ubuntu0.1) jammy; urgency=medium
* New upstream point release (LP: #1969482):
- bufixes
- performance fixes
- v5.17 & v5.18 compat
- scrub services
- fix fallocate LP: #1969247
The update fixed my lxd containers using the same method from above on
my zfs jammy system. However it did not fix the base system itself. When
I try to install mysql-server or run the fallocate test it still fails.
Are there additional steps, other than just apt updating and upgrading,
that I or
Hello Lena, or anyone else affected,
Accepted zfs-linux into jammy-proposed. The package will build now and
be available at https://launchpad.net/ubuntu/+source/zfs-
linux/2.1.4-0ubuntu0.1 in a few hours, and then in the -proposed
repository.
Please help us by testing this new package. See
** Changed in: zfs-linux (Ubuntu)
Status: Triaged => In Progress
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces
** Changed in: zfs
Status: New => Fix Released
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on
zfs-dkms ended up working for me using LXD with the following test case:
sudo add-apt-repository ppa:ci-train-ppa-service/4831
sudo apt update
sudo apt upgrade
sudo apt install zfs-dkms
lxd init
lxc launch images:ubuntu/focal test-mysql
lxc exec test-mysql bash
# apt update && apt dist-upgrade
I've picked up the PR into a bileto ppa
https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/4831
it seems to work for me correctly. If one has MOK dkms key enrolled or
secureboot disabled, one can use zfs-dkms package from the above ppa as
a solution. Please test it and let me know.
I
** Changed in: mysql-8.0 (Ubuntu)
Status: Triaged => Invalid
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces
yeap that's bad.
** Changed in: zfs-linux (Ubuntu)
Status: Confirmed => Triaged
** Changed in: zfs-linux (Ubuntu)
Importance: Wishlist => High
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
On jammy, it works on btrfs:
root@nsn7:/mnt# rm -f foo.img; touch foo.img; fallocate -z -l 10M foo.img; ls
-lah foo.img
-rw-r--r-- 1 root root 10M abr 19 10:46 foo.img
root@nsn7:/mnt# mount -t btrfs
/btrfs.img on /mnt type btrfs
(rw,relatime,ssd,space_cache=v2,subvolid=5,subvol=/)
--
You
I applied zfs_fallocate_reserve_percent=0 to the kernel module,
confirmed it was set, but I still see the same incorrect behavior with
fallocate:
root@nsn7:/tank/test# rm -f foo.img; touch foo.img; fallocate -z -l 10M
foo.img; ls -lah foo.img
-rw-r--r-- 1 root root 0 abr 19 10:35 foo.img
Impish kernel on zfs:
andreas@nsnx:~$ rm foo.img
andreas@nsnx:~$ touch foo.img
andreas@nsnx:~$ fallocate -z -l 10M foo.img
fallocate: fallocate failed: Operation not supported
Jammy kernel on zfs:
root@j1:~# touch foo.img
root@j1:~# fallocate -z -l 10M foo.img
root@j1:~# ll foo.img
The impish kernel returns "not supported" when fallocate is used in the
way mysql does it. That makes mysql aware of the problem, and switch
strategies, and it works.
The jammy kernel is saying all is fine with the call and its results,
but it isn't. Mysql, or any userspace, shouldn't be second
imho, mysql should still gain zfs specific knowledge that doing
fallocate on CoW filesystems doesn't do what one expects it to do.
** Changed in: mysql-8.0 (Ubuntu)
Status: Invalid => Triaged
** Changed in: zfs-linux (Ubuntu)
Importance: Undecided => Wishlist
--
You received this bug
However there is now also this:
https://github.com/openzfs/zfs/pull/13338/files
which seems to indicate that my / zfs understanding of the fallocate
kernel flags are not what is done on other filesystems (and expected by
stuff).
If needed, we can look into pulling in
given how zfs works, to me this is normal and expected.
Also this is expected on btrfs, and on xfs with cow turned on.
zfs is copy on write filesystem, thus calling fallocate to "reserve" a
large amount of free space doesn't make sense. As the created file is
immutable in zfs, and any writes to
** Changed in: zfs
Status: Unknown => New
--
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to zfs-linux in Ubuntu.
https://bugs.launchpad.net/bugs/1969247
Title:
fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on zfs in
** Summary changed:
- fallocate produces zero-size files on zfs in Jammy
+ fallocate with FALLOC_FL_ZERO_RANGE produces zero-size files on zfs in Jammy
** Description changed:
- I came across this error when testing various mysql setups in an LXD
- container and managed to reproduce it
43 matches
Mail list logo
