Bug 718208

Summary: fence_ilo fails silently when pyOpenSSL is not installed
Product: Red Hat Enterprise Linux 5 Reporter: Marek Grac <mgrac>
Component: cmanAssignee: Marek Grac <mgrac>
Status: CLOSED WONTFIX QA Contact: Cluster QE <mspqa-list>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 5.6CC: cluster-maint, edamato, mgrac, mkelly, zidek
Target Milestone: rc   
Target Release: ---   
Hardware: x86_64   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: 698002 Environment:
Last Closed: 2011-08-05 14:41:40 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On: 698002    
Bug Blocks:    

Description Marek Grac 2011-07-01 12:01:16 UTC
+++ This bug was initially created as a clone of Bug #698002 +++

Description of problem:

The cman RPM does not depend on pyOpenSSL, and fence_ilo simply fails with "Unable to connect/login to fencing device", regardless if '-v' or '-D /tmp/foo' are passed. This makes it very difficult to sort out what is wrong.

It is true that the fence_ilo man page indicates that this is a required package, but it is easily missed when people skip the DESCRIPTION section and go right to reading the switches.

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

cman-2.0.115-68.el5.x86_64
fence_ilo: FENCE_RELEASE_NAME="2.0.115";

How reproducible:

100%

Steps to Reproduce:
1. Uninstall pyOpenSSL
2. Run 'fence_ilo -a <ip> -l <user> -p <pass> -o reboot
3.
  
Actual results:

Fails with a generic error message

Expected results:

Report that a dependent package is missing.

Additional info:

--- Additional comment from zidek on 2011-05-24 06:52:05 EDT ---

cman should be dependent on pyOpenSSL. Today I run into huge problems after cluster split brain condition with this error.

Comment 1 Madison Kelly 2011-07-01 16:03:02 UTC
If I recall correctly, this was a design decision meant to avoid a large number of dependencies that wouldn't be needed for most implementations. However, I think this has changed in RHEL 6. This said, I am a user, so I could be wrong on this. If a developer replies differently, disregard this. :)

Comment 2 Lon Hohberger 2011-08-05 14:41:40 UTC
Digimer is correct; the idea was to reduce dependencies.  This is resolved in RHEL-6, where we did no such thing in the fence-agents package.