Bug 86527

Summary: up2date reports "package xxx does not have a valid GPG signature" when the error is really "Connection reset by peer"
Product: [Retired] Red Hat Linux Reporter: Tim Bingham <tim.bingham>
Component: up2dateAssignee: Adrian Likins <alikins>
Status: CLOSED NOTABUG QA Contact:
Severity: medium Docs Contact:
Priority: medium    
Version: 8.0CC: gafton, mihai.ibanescu
Target Milestone: ---   
Target Release: ---   
Hardware: i686   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2004-08-20 20:25:17 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Tim Bingham 2003-03-25 03:49:44 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20021003

Description of problem:
Running up2date to download a large package over a slow (56k) link gets the
following error in a dialog box ...

"The package kdelibs-3.0.3-8.3 does not have a valid GPG signature.
It has been tampered with or corrupted.  Continue?"

The error occurs during download after a portion of the package has been received.

No other errors are apparent unless one starts up2date from a shell window ..

 
# up2date
error: rpmts_HdrFromFdno: MD5 digest: BAD
Expected(f65fe4c6774d31896625bf3a5d4037b1) != (a79cc8f7fff7ca61d41cb2076279e81e)
SSL exception (104, 'Connection reset by peer')
SSL exception (-1, 'Unexpected EOF')
error: rpmts_HdrFromFdno: V3 DSA signature: BAD, key ID db42a60e
#

Note that the first "rpmts_HdrFromFdno" error is printed immediately (before the
download has started) while the other errors happen right before the "package
xxx does not have a valid GPG signature" dialog box pops up.



Version-Release number of selected component (if applicable):
up2date-3.0.7-1

How reproducible:
Sometimes

Steps to Reproduce:
1.Run up2date from a shell window to download a large package over a slow (56k) link
2.The download fails with "package xxx does not have a valid GPG signature"
3.Note "Connection reset by peer" error in shell window.
    

Actual Results:  An incorrect error ("package xxx does not have a valid GPG
signature") was reported and the package download was aborted.


Expected Results:  The correct error ("Connection reset by peer") is reported.
The connection is reestablished and the download is resumed where it left off.

Additional info:

This bug causes confusion because unless up2date is started from the shell the
real cause of the error is lost.

Comment 1 RbR 2004-03-29 09:37:52 UTC
A couple of examples: 
 
The package iptables-1.2.9-1.0 does not have a valid GPG signature. 
It has been tampered with or corrupted.  Continue? 
 
The package mc-4.6.0-8.4 does not have a valid GPG signature. 
It has been tampered with or corrupted.  Continue? 
 
Option Yes is needed; otherwise the up2date procedure is completely 
cancelled 

Comment 2 Adrian Likins 2004-08-20 20:25:17 UTC
networking issues that cause package corruption should
be better detected in current  versions.

Either way, up2date is doing the correct thing by warning.

Comment 3 Andre Robatino 2004-08-20 22:31:18 UTC
  The warning message is wrong.  The package has NOT been tampered
with or corrupted; rather, it has been truncated due to the server
dropping the connection (the download can be finished successfully
using wget).  The correct behavior would be for up2date to check
whether the file size is correct first, and if it's too short, it
should say that the downloaded file is truncated and to use some other
tool capable of resuming downloads to finish it.  In any case, the
problem still exists with FC2.  Using mirrors (which is the only
improvement since 8.0) is NOT a fix, it simply makes the bug less
likely to manifest.  (I still experience it now and then.)  When
up2date can resume an aborted download (or at least if it gives a
correct warning message), then it'll be fixed.  Please reopen this bug
(and any related bug reports closed for the same reason, for example
bug #85808 which was reported as closed in email, though the web page
itself isn't updated yet).