Bug 2088731 (F37Workstationraw-xzaarch64Oversize) - Fedora 37: Workstation raw-xz aarch64 image exceeds maximum size
Summary: Fedora 37: Workstation raw-xz aarch64 image exceeds maximum size
Keywords:
Status: CLOSED RAWHIDE
Alias: F37Workstationraw-xzaarch64Oversize
Product: Fedora
Classification: Fedora
Component: distribution
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Chris Murphy
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks: FedoraOversizeTracker F37BetaBlocker F37FinalBlocker
TreeView+ depends on / blocked
 
Reported: 2022-05-20 10:35 UTC by Fedora QA Tools SIG
Modified: 2022-07-31 15:20 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2022-06-24 23:40:17 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Fedora QA Tools SIG 2022-05-20 10:35:08 UTC
Workstation raw-xz aarch64 image https://kojipkgs.fedoraproject.org/compose/rawhide/Fedora-Rawhide-20220520.n.0/compose/Workstation/aarch64/images/Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw.xz from compose Fedora-Rawhide-20220520.n.0 is 4040775216 bytes, exceeding the maximum size 4000000000. Canonical maximum sizes can be found at https://fedoraproject.org/wiki/Releases/37/Spins and https://fedoraproject.org/wiki/Releases/37/ReleaseBlocking . This check is run by the 'relval' tool, which has its own list of maximum sizes derived from those pages. If the maximum size used for this comparison is wrong, please add a comment and file a bug against relval at https://pagure.io/fedora-qa/relval/issues and it will be corrected. If you believe the canonical maximum size for an image should be changed, please follow the appropriate process before filing a relval bug.

Comment 1 Ben Cotton 2022-05-23 13:49:51 UTC
Assigning to Chris as Workstation co-chair since Allan is out for a few weeks.

Comment 2 Chris Murphy 2022-05-23 22:08:07 UTC
This seems really familiar...I suspect lack of FITRIM being issued or passed down, thus the image isn't sparse.

3946072 -rw-r--r--. 1 chris chris 4040775216 May 23 16:41 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw.xz
...
3.8G -rw-r--r--. 1 chris chris 3.8G May 23 16:41 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw.xz

unxz it

5588764 -rw-r--r--. 1 chris chris 13958643712 May 23 16:41 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw
...
5.4G -rw-r--r--. 1 chris chris  13G May 23 16:41 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw

$ sudo losetup /dev/loop0 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw
$ sudo kpartx -a /dev/loop0
$ sudo mount /dev/mapper/loop0p1 /mnt/1
$ sudo mount /dev/mapper/loop0p2 /mnt/2
$ sudo mount /dev/mapper/loop0p3 /mnt/3
$ mount | grep loop
/dev/mapper/loop0p1 on /mnt/1 type vfat (rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=ascii,shortname=mixed,errors=remount-ro)
/dev/mapper/loop0p2 on /mnt/2 type ext4 (rw,relatime,seclabel)
/dev/mapper/loop0p3 on /mnt/3 type btrfs (rw,relatime,seclabel,ssd,space_cache=v2,subvolid=5,subvol=/)
$ sudo fstrim -v /mnt/1
/mnt/1: 568.3 MiB (595902464 bytes) trimmed
$ ls -ls
5588740 -rw-r--r--. 1 chris chris 13958643712 May 23 17:20 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw
$ sudo fstrim -v /mnt/2
/mnt/2: 833.5 MiB (874020864 bytes) trimmed
$ ls -ls
5492040 -rw-r--r--. 1 chris chris 13958643712 May 23 17:20 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw
[chris@fnuc fedora37]$ sudo fstrim -v /mnt/3
/mnt/3: 8.5 GiB (9149927424 bytes) trimmed
$ ls -ls
3202480 -rw-r--r--. 1 chris chris 13958643712 May 23 17:20 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw

Nearly 2.2G can be trimmed away. And if I xz this image up and compare...


2084228 -rw-r--r--. 1 chris chris 2134248116 May 23 17:32 Fedora-Workstation-Rawhide-20220520.n.0.aarch64.raw.xz

3946072-2084228=1861844 or ~1.8G

So > 2G of garbage is being needlessly processed with xz. And ~1.8G needlessly downloaded by everyone. And presumably other Oz produced images are affected, and nightly.

Comment 3 Chris Murphy 2022-05-23 22:23:08 UTC
OK so @pbrobinson told me that this is basically using Anaconda in a libvirt VM.

There should be a lorax update soon, that will result in fstrim being issued to all file systems before the final unmount: https://bugzilla.redhat.com/show_bug.cgi?id=1971186

This also needs Fedora releng to work on this issue: https://pagure.io/releng/issue/10801

Comment 4 Chris Murphy 2022-05-23 22:23:39 UTC
Oops.

Comment 5 Adam Williamson 2022-06-24 23:40:17 UTC
It looks like this is undersize on recent composes. Today's image is 3784697224 . Sounds like more could be saved, but we can close this.

Comment 6 Peter Robinson 2022-06-25 08:38:20 UTC
(In reply to Adam Williamson from comment #5)
> It looks like this is undersize on recent composes. Today's image is
> 3784697224 . Sounds like more could be saved, but we can close this.

I suspect that's from the recent reductions I've made in linux-firmware

Comment 7 Adam Williamson 2022-06-25 16:13:05 UTC
probably more than that, as this was 500M of savings. Not sure what caused it, though. Note the netinsts are oversize again :( I'll look into that next week.


Note You need to log in before you can comment on or make changes to this bug.