during installation/update of huge packagesets from the download.opensuse.org mirror network I often get connection errors like:
Could not connect to host
After hitting retry, update process continues.
This is definitely not a problem with my internet connection.
It would also be great to have some additional information how the mirror network works. Is every package fetched from a different mirror, or is it connection based e.g.?
Okay, that sounds as if every request is handled separately, so it may happen that every single package is downloaded from a different mirror (correct me if I am wrong).
For me, it is still not clear what causes the timeouts. Is it because of broken mirrors? Shouldn’t the periodic check detect and mark/flag them?
How can I debug this issue?
Some of the cases of ‘timeouts’ may not necessarily appear to all users, some of them may be ISP routing related (as such only affecting certain % of users) and/or may have to do with periodic issues at their end which are only intermittent and again may only last a few seconds / minutes.
Maybe you are right but the situation is very unsatisfying (and never had these problems with other networks, e.g. Fedora).
ATM I have to monitor updates to click ‘retry’ after a timeout.
Is there a way to define the timeout value or set something like auto-retry?
Also I would prefer a session-based mirror network instead of ‘per request’.