Red Hat Bugzilla – Bug 211117
Anaconda traceback + forced reboot when file temporarily unavailable
Last modified: 2007-11-30 17:11:46 EST
Description of problem:
Just tried a HTTP install of rawhide. I had done the mirroring wrong -> the RPMS
directory was missing. It gave me a Cancel/Retry option, after fixing the
directory structure and pressing Retry it did try again, and then gave the
traceback attached. No way of continuing the install.
Similar error occurred when doing a FTP install off a FTP server, and the server
started responding too many connections (10) from your IP address after a while.
No way to recover from there either, even though retrying the operation after a
while would have been fine.
Created attachment 138684 [details]
Applied a fix for this.
Updates image available:
211117-updates.img fixed the problem, thanks!
I retried the scenario, this time by removing the network cable in the middle of
the install and waiting a long time for a timeout. Finally the reboot/retry
dialog came up, I pressed it a few times (first press didn't seem to do
anything, after the second it seems like it actually started retrying), then
reconnected the cable and the install is now happily continuing...
(If someone wants to further test these kinds of things, I recommend a local
mirror + then just removing files or whatnot to trigger, the timeout takes ages
with the "remove cable" approach, but it eventually does. Sensible behaviour,
except when you want to test the failure code ;) )
This bug appears to be fixed in FC6.