Red Hat Bugzilla – Bug 35713
Errors raised using SSL with python (up2date)
Last modified: 2015-01-07 18:45:05 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.2.17-14cl i686; en-US; 0.8.1)
After an update in up2date(to version v2.1.7), the program can't connect to
the server because an "SSL_connect error". Stunnel was updated too.
Stunnel to version 3.10-2.
Steps to Reproduce:
1.update up2date to version 2.1.7, build up2date-2.1.7-1
2.update stunnel to version 3.10, build stunnel-3.10-2
3.run up2date -p (or any argument that requires secure connection)
Actual Results: Updating package profile...
Traceback (innermost last):
File "/usr/sbin/up2date", line 455, in ?
File "/usr/sbin/up2date", line 350, in main
File "/usr/share/rhn/up2date/up2date.py", line 702, in updatePackageProfile
up2date.CommunicationError: Error communicating with server. The message was:
Expected Results: Should performed an update of my profile in the RHN server.
Is this error repeatable? Does it work if you aren't using stunnel?
when I say repeatable, I mean is it repeatable today? If it was a transient
server side issue, it has probably been addressed. If not I need to probe
also, what version of python is installed?
The bug is repeatable any time. I thought was an server problem too, but no.
The python packages are:
the up2date program worked with the standard installation of RH 7 for alpha
system. But in that time was impossible to update de files cause up2date
couldn't retreive the files because a missing part of the rpm name (alpha). In
that time an up2date update was already available and I updated the files. Since
the update I get this error (SSL error) message.
This is indeed an issue for alphas. I am able to replicate it in the test lab
and reported it to the openssh guys as an issue to take a look at. In my
testing, I have found that all SSL activity on the alpha is doing this, so this
is definitely not an up2date issue specifically.
Nalin, Cristian: can you look into the python/openssl combo on alpha, and
determine what is going wrong?
This was recently fixed (caser of memcpy() being used in a overlapping memory
Nothing to see here anymore