Fedora Account System
Red Hat Associate
Red Hat Customer
Server boot aarch64 image https://kojipkgs.fedoraproject.org/compose/branched/Fedora-45-20260927.n.0/compose/Server/aarch64/iso/Fedora-Server-netinst-aarch64-45-20260927.n.0.iso from compose Fedora-45-20260927.n.0 is 1435201536 bytes, exceeding the maximum size 1400000000. Canonical maximum sizes can be found at https://docs.fedoraproject.org/en-US/releases/f45/spins/ and https://docs.fedoraproject.org/en-US/releases/f45/blocking/ . 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.
*** Bug 2544630 has been marked as a duplicate of this bug. ***
Marking accepted, as Kamil explained on the other bug, this is an automatic blocker.
So, let's see. Compared to F44 (where the image was ~1.2G): Firmware is 70M bigger - 561M vs. 494M. Kernel modules are 27M bigger - 201M vs. 174M. The 70M difference in firmware, I think, is duplicated in the PXE boot initramfs - images/pxeboot/initrd.img - which is 294M vs. 224M. That's 167M total already. Uncompressed, but kernel drivers and firmwares are all I think compressed already so we don't gain anything by recompressing them. I guess that's a lot of the difference. About half the difference in firmware is qualcomm adding more stuff: qcom is 169M vs. 133M. iwlwifi is 12M bigger, 89M vs. 78M. ath12k gained too, 15M vs. 8.8M, and amdgpu gained 2M. mediatek is 37M vs. 26M (mt7927 appeared and mt7966 nearly doubled in size). I think that's most of the diff. Peter, is there anything we can do about any of that? Outside of that, we gained 12M in /usr/bin. half of that is /usr/bin/bootc: it was 9.1M in F44, it's 15M in F45. Colin, is that necessary? Most of the rest, I think, is due to adding stratis: we have stratis-predict-usage and stratis-decode-dm which weren't there before. It seems blivet actually uses stratis-predict-usage, but it doesn't use stratis-decode-dm AFAICS. We could trim this a bit with lorax for now but it seems like wasted effort when we'll likely switch to image-builder in future, and maybe stratis itself might call stratis-decode-dm, I dunno... openssl stuff accounts for a bit more. In F44 we have /usr/lib64/ossl-modules which is 1.6M; in F45 it's 4.6M and we *also* have /usr/lib64/ossl-modules-3 which is 1.6M. So I guess we have both openssl3 and 4 in the installer env (unfortunate) and 4's modules are rather bigger than 3's. There's a /usr/lib64/python3.15/profiling/sampling dir which didn't exist on Python 3.14 in F44 and takes 1.1M. /usr/lib64/systemd went from 7.7M to 8.8M. We're dragging around /usr/share/nss/unsupported-tools which I bet we don't use. (Also we have *three* crypto libraries in the installer environment?! No, scratch that, gnutls is there too. We have four crypto libraries in the installer environment.) It's really just a bunch of things...
Server boot aarch64 image https://kojipkgs.fedoraproject.org/compose/branched/Fedora-45-20261003.n.0/compose/Server/aarch64/iso/Fedora-Server-netinst-aarch64-45-20261003.n.0.iso from compose Fedora-45-20261003.n.0 is 1435121664 bytes, exceeding the maximum size 1400000000.
Server boot aarch64 image https://kojipkgs.fedoraproject.org/compose/branched/Fedora-45-20261009.n.0/compose/Server/aarch64/iso/Fedora-Server-netinst-aarch64-45-20261009.n.0.iso from compose Fedora-45-20261009.n.0 is 1436108800 bytes, exceeding the maximum size 1400000000.