https://bugzilla.novell.com/show_bug.cgi?id=425035
User poeml@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=425035#c4
Peter Poeml
Failures sometimes happen because download.opensuse.org is a cluster with several physical servers in it. Some of them sometimes are heavily loaded thus causing timeout. zypper's request distribution is round-robin afaik, thus next attempt leads to another server in cluster that is more or less free/responsive. Sometimes (if using factory) some of servers may contain corrupted packages, most likely when new packages are being propagated thru all mirrors. This is just IMHO - i judge as what i see how it works. Maybe zypper should have tunable retry count initially set to number of servers in cluster?
The real problem is that the bunch of mirrors that download.opensuse.org redirects to provide unpredictable service. Both the client side and the server side can help to deal with this. Server side: ============ The most efficient way to deal with it is to help them improve their service, and help us (openSUSE) to make better use of mirrors by knowing them better. Therefore, https://features.opensuse.org/?rm=feature_show&id=305320 is what would be most useful at the moment. To read more about the architecture, I recommend http://mirrorbrain.org/ and http://en.opensuse.org/Build_Service/Redirector Client side: ============ In addition, the implementation of http://en.opensuse.org/Libzypp/Failover is coming. The concept is superior to simple retries as suggested here (which is not a bad idea as such, of course). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.