Qubes Updater giving me too many problems

User Support General
1

Many years ago I switched from manually updating each qube (using apt or dnf) to using the Qubes Updater GUI because I read that the devs recommended this. As of today I’m switching back as I’m having too many problems with the Updater. I’ll list them.

  1. For a long time now 1 qube will not update with the updater and I have to go in and manually do it every time. First time I do apt update it falls to hit anything, then I run it second time and everything works fine. No idea why.
  2. Updater has begun crashing mid update. This has happened a few times now where I come back to check on my updates and the updater has closed. I start it back up and see that only some of the qubes had updated. I run it again and many of them show Already downloaded over and over indicating that it had gotten to a certain point of downloading updates for multiple qubes but stopped before installing them.
  3. Updater is much slower than manually updating. I will list some of the messages below that seem to be behind this slowness. [note I run updates over Tor]

Removed 0 files, 0 directories (total of 0 B). 0 errors occurred.
Already downloaded
Already downloaded
Already downloaded
Already downloaded
Already downloaded
Already downloaded
Already downloaded
Already downloaded
Fetching 58 packages [688.10 MiB]
Fetching None failure (nvidia-gpu-firmware-0:20261019-1.fc43.noarch) Interrupted by header callback: Inconsistent server data, reported file Content-Length: 57820351, repository metadata states file length: 104080575 (please report to repository maintainer)
Fetching None failure (linux-firmware-0:20261019-1.fc43.noarch) Interrupted by header callback: Inconsistent server data, reported file Content-Length: 13082764, repository metadata states file length: 36626572 (please report to repository maintainer)
…
Fetching None failure (linux-firmware-0:20261019-1.fc43.noarch) Curl error (56): Failure when receiving data from the peer for https://southfront.mm.fcix.net/fedora/linux/updates/43/Everything/x86_64/Packages/l/linux-firmware-20261019-1.fc43.noarch.rpm [OpenSSL SSL_read: OpenSSL/3.5.9: error:0A000126:SSL routines::unexpected eof while reading, errno 0]
Fetching None failure (nvidia-gpu-firmware-0:20261019-1.fc43.noarch) Curl error (56): Failure when receiving data from the peer for https://southfront.mm.fcix.net/fedora/linux/updates/43/Everything/x86_64/Packages/n/nvidia-gpu-firmware-20261019-1.fc43.noarch.rpm [OpenSSL SSL_read: OpenSSL/3.5.9: error:0A000126:SSL routines::unexpected eof while reading, errno 0]
…

2

Is this an old template copied from Debian or Kicksecure? Do you keep maintaining those older copied templates, replacing them with fresh copies from time to time?

What update template are you using, and which template does it base on? is Tor routed through the VPN?
I believe points 2 and 3 are somehow related to point 1, and that they could be resolved or improved by renewing your templates.

In short, I encountered the same issue and resolved it by performing updates through the CLI (dom0 CLI updater) and by replacing the templates with fresh builds.

3

It’s a StandAlone based on whonix-gateway that has probably been updated from a previous version of whonix once or twice. It wouldn’t be hard to rebuild it from scratch, but has been easier to just manually update to date.

Default update proxy is set to sys-whonix in Qubes OS Global Config, which is based on the whonix-gateway-18 template. This template should have been in stalled new when whonix 18 came out, so not as old as the one above.

4

Please check whether you can update this template using the Qubes Template Manager.