Due to an influx of spam, we have had to temporarily disable account registrations. Please write an email to accountsupport@archlinux.org, with your desired username, if you want to get access. Sorry for the inconvenience.
Reflector seems to have problems fetching mirrors. I only get one mirror from china. Maybe it has something to do with completion rate, but inside the EndeavourOS forum some people also have this problem (I'm using vanilla arch).
Especially annoying when booting from an arch iso for installing arch.
package version: 2023-1
Steps to replicate:
Just run reflector.
Edited
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Child items ...
Show closed items
Linked items 0
Link issues together to show that they're related or that one is blocking others.
Learn more.
Something definitely seems amiss.. but reflector hasn't been updated in quite a while, so probably something has changed at the Arch end. cc @torxed@pitastrudl for visibility. Any ideas?
The problem was in percentage of fullness of the mirrors - only three of them were 100% full yesterday and only one of them has https protocol. Reflector only works with the mirrors are 100% full by default, so it only showed one server.
As a solution, the '--completion-percent 98' option can be added to the Arch install script, but it can lead to a more serious problem - partial updates
Even if it only showed one mirror, reflector.service should have ended up in a "completed" state right? Or what happened? Because archinstall monitors the service, waiting for it to be done, dead or similar states.
All I can say is that I've gotten multiple pings about it too, where people blame archinstall *
(the symptoms just show up there)*.
Runnig reflector --verbose manually shows nothing and it hangs.
Spamming Ctrl-C basically does nothing either.
The endpoint https://archlinux.org/mirrors/status/json/ appears to be working.
At the very least, Reflector could implement signal handling so outside influences could trigger signals (even via systemd that could be useful).
Further, I don't know how that xyne.dev "source code" repo works. But the issue appears to hit around line ~122, in Reflector.py(/usr/lib/Python3.11/Site-Packages/Reflector.py):
And that code works if run outside of the ISO.
But inside the ISO it hangs.
The root cause appears to be (in my test) the DNS lookup of archlinux.org, which is why the timeout=5 isn't effective.
Looking at journalctl -f -u systemd-resolved.service I got a bunch of:
Using degraded feature set UDP instead of TCP for DNS server <ip of dns server>Using degraded feature set TCP instead of UDP for DNS server <ip of dns server>
I noted that we wrote at the exact same time, and hence why I edited the comment.
However I'm wondering if the the default conf of resolved and the DNS lookup issue happens for others too. As I constantly get issues about reflector and hanging things.
I always refer to "git gud, fix your network" but perhaps there's some improvement to be made here too?
Hey, just wanted to drop this bit of information. Seems that one of the location that archweb checks from (London located server) was unavailable for around 2 hours give or take. In that time all the checks from that location failed, making it seem like a mirror is unavailable which then in turn caused the Completion % to drop thus causing reflector to stop working. You can verify this by going to any server log history and looking at around 4AM on the 18th of January. You'll see pretty much everywhere that there was an issue. This data point will not be visible anymore in a couple of days as it rotates through.