Red Hat Bugzilla – Bug 390671
ftp servers without resume not supported when retrying partly downloaded RPM
Last modified: 2008-01-27 14:16:25 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; fr; rv:18.104.22.168) Gecko/20071019 Fedora/22.214.171.124-1.fc7 Firefox/126.96.36.199
Description of problem:
On a server that doesn't support "resume", after having been interrupted through a download, when restarting the download, Smart is keeping asking for resume to the server with no option to modify this behavior, then the user can't download the file.
The only mean to download the RPM is to delete the partial download in the Smart's cache directory (/var/lib/smart/packages/).
Then Smart is downloading the file entirely from the beginning of that file.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. in Smart GUI add the ftp Fedora repo (RPM Metadata): ftp://ftp.proxad.net/mirrors/fedora.redhat.com/fedora/linux/releases/8/Everything/i386/os
2. enable and update this new channel, then choose a big RPM to download (e.g. openarena-data = 266MB) and click on "Apply changes" button
3. cut your network connection after a few percent done, wait for Smart to display an error message
4. re-establish your network connection and retry "Apply Changes" button
5. Smart will pop up a message telling you the server doesn't know this command (error 500)
6. then you will have no choice but open a terminal and delete the partial download in the Smart cache (reboot, restart Smart, etc. won't work) to be able to download this RPM
Smart keeps on trying to "resume" the partial download leaving you no choice but open a terminal, login as root and delete the partial download in the Smart cache (reboot, restart Smart, etc. won't work).
Then you will be able to download this RPM entirely.
on error 500, Smart may ask user if he wants to start downloading the file all over again, or keep trying to resume the incomplete download.
Or more simply after some failed retries, just try to download the file all over again, ignoring the already downloaded part of the file.
Or add a button to cleear the partially downloaded files from the Smart's cache.
But it shouldn't try to resume forever without an option to clear the cache or try to download the file entirely (ignoring already downloaded part).
This is an ugly bug, I filed it upstream in http://tracker.labix.org/issue344
Feedback: The bug has been fixed upstream, but there is no new release yet.