Mirrors timing out in docker #116
Comments
|
Additionally on the Truenas Docker I get this if I set a predefined server |
|
My first guess would be that the mirror selection does not work well for you and flexo is therefore choosing mirrors with a too high latency. You don't have anything unusual configured regarding networking, like a VPN or so? Also, on which continent are you located? In the past, some users from Australia had issues because of the latencies from Australia to mirrors located in Europe and elsewhere. Could you please try the following:
If that still doesn't work, please post the output of the following command, would be interesting to see how fast your connection is: curl -w @- -o /dev/null -s "https://mirror.moson.org/arch/core/os/x86_64/core.db" <<'EOF'
time_namelookup: %{time_namelookup}\n
time_connect: %{time_connect}\n
time_appconnect: %{time_appconnect}\n
time_pretransfer: %{time_pretransfer}\n
time_redirect: %{time_redirect}\n
time_starttransfer: %{time_starttransfer}\n
----------\n
time_total: %{time_total}\n
EOF |
|
Trying this on my laptop and it works file. Trying this on my server which is a TrueNAS Scale x86_64 which uses kubernetes. Flexo also exits because it can't find any non timed out mirror. Increasing max score to 10 and timeout to 30 seconds doesn't change anything. Here's the output with moson.org |
So it takes more than 5 seconds until name lookup has completed, that is very unusual. This is how it looks for me: It's possible that flexo does not behave quite as expected under less-than-ideal networking conditions, but my impression is that some networking strangeness in your Kubernetes setup seems to be at least a contributing factor. Have you verified that the settings are also applied correctly for Flexo? I'm not an expert on Kubernetes, but with plain docker, here's one example to set the During startup, you should then see something like the following to show which settings are applied (this requires a log level of DEBUG or comparable): |
|
What you say makes sense, the time takes less long on other devices. You can see the settings i ran in my original post. |
|
Using the Don't know what we can learn from this to improve flexo since these issues only came up from a bad DNS config. |
Glad to hear that!
I might try to check if I can reproduce this issue somehow, but as long as it works fine for users with no unusual network settings, I won't invest too much time into it. There are other areas at flexo that have higher priority. |
Describe the bug
When running the flexo docker, no mirrors are reachable and they time out. Increasing the time out duration does nothing.
Running on two different machines (Arch Linux x86_64, TrueNAS Scale cobia x86_64) gives the same result: Locally, the mirrors are reachable, but in the docker they time out.
Starting up an arch linux docker and installing flexo from the AUR results in timeouts as well.
Running curl inside of the docker works
Installation method
Docker
Distribution
linux/amd64
Version
digest e9f3ba02b33a
Log
Running
docker run -it --rm -e RUST_LOG=trace nroi/flexo:latestThe text was updated successfully, but these errors were encountered: