You are not logged in.
I have a headless server and it is showing some strange issues since about 1-2 years. Basically, it is extremely difficult to do system update, because the process constantly times out either on probing repositories:
:: Synchronising package databases...
archzfs is up to date
zfs-linux is up to date
zfs-linux-lts is up to date
core is up to date
extra is up to date
community is up to date
error: failed retrieving file 'archzfs.db' from archzfs.com : Resolving timed out after 10001 milliseconds
error: failed retrieving file 'archzfs.db' from mirror.sum7.eu : Failed to connect to mirror.sum7.eu port 80 after 10000 ms: Timeout was reached
:: error: GET https://www.archlinux.org/feeds/news/
:: error: urllib: [Errno 101] Network is unreachable
and while downloading packages.
Total Download Size: 442.84 MiB
Total Installed Size: 763.27 MiB
Net Upgrade Size: -21.89 MiB
:: Proceed with installation? [Y/n] Y
:: Retrieving packages...
linux-firmware-20240409.1addd7dc-1-any 222.9 MiB 51.3 MiB/s 00:04 [#############################################] 100%
linux-lts-6.6.27-1-x86_64 128.8 MiB 1122 KiB/s 01:57 [#############################################] 100%
linux-lts-headers-6.6.27-1-x86_64.pkg.tar.zst failed to download
Total ( 3/41) 442.8 MiB 3.35 MiB/s 02:12 [#############################################] 100%
error: failed retrieving file 'linux-lts-headers-6.6.27-1-x86_64.pkg.tar.zst' from kernels.archzfs.com : Operation too slow. Less than 1 bytes/sec transferred the last 10 seconds
The transfer will be fine for several seconds, and then suddenly start slowing down to zero. Then I restart the process, it times out again. And again. And again. And then it works for another 1-20 packages, and times out again. You get the picture.
It takes 10-30 attempts to restart the process to complete the update. What's strange is that a laptop on the same connection (also with Arch) does not have such issues.
This is not the matter of choosing appropriate mirrors.
I have no idea what can be causing this.
Last edited by Lockheed (2024-04-16 22:06:34)
Offline
This is not the matter of choosing appropriate mirrors.
Yet the errors you show are only w/ kernels.archzfs.com ?
mirror.sum7.eu is 404, archzfs.com 302 to github, kernels.archzfs.com/archzfs is 404…
mirror.biocrafting.net seems up, does that provide better results?
Offline
I assure you kernels.archzfs.com is just an example. The issue happens randomly for every repo.
And since I started running OpenWRT in LXD on the same host, I think can see the same problem when I try to update its packages. Strangely, my SSH or http connection to this remote host does not experience any slowdowns during those timeouts/lack of connectivity.
Offline
linux-firmware-20240409.1addd7dc-1-any 222.9 MiB 51.3 MiB/s 00:04 [#############################################] 100%
linux-lts-6.6.27-1-x86_64 128.8 MiB 1122 KiB/s 01:57 [#############################################] 100%
linux-lts-headers-6.6.27-1-x86_64.pkg.tar.zst failed to download
Total ( 3/41) 442.8 MiB 3.35 MiB/s 02:12 [#############################################] 100%
You're starting > 50MB/s, then break down to 1MB/s before the connection collapses after 442.8MB throughput - that's enough for a veeeeeeery long ssh session (if the pattern is somewhat reflective of the general condition) but it would probably be nice to run a speedtest from that host to see whether that's pacman/mirror specific or a general network issue
Get ABBS abstracts some linode downloads and also check the system journal whether the connection loss is somehow locally reflected (carrier loss, driver crash etcetc.)
Offline
The speed goes down to zero for minutes and the SSH session and graphical remote desktop connection remains unbroken.
Anyway, I just discovered that at some point the ISP changed the internal IP of my upstream router to the same IP as the internal IP of my downstream home router and I suspect it might be causing the issue. The laptop client connects to the WiFi of the upstream router so would not be affected. I will report back after resolving the conflict.
Offline
Well, resolving the IP conflict seems to have resolved the issue. Thanks for the attention.
Offline