Red Hat Bugzilla – Bug 1236599
For SCVMM hypervisor, verifying host credentials throws EPIPE
Last modified: 2016-08-16 10:42:55 EDT
Description of problem: SCVMM hypervisor verifying credentials throws EPIPE. Version-Release number of selected component (if applicable): 5.4.0.5.20150605150206_7daa1a8 How reproducible: 100% Steps to Reproduce: 1. Define Microsoft infrastructure provider 2. Refresh Relationships and Power States 3. For a discovered host, verify credentials Actual results: Unexpected response returned from Cloud/Infrastructure Provider, see log for details Expected results: Verified credentials Additional info: evm.log: [----] I, [2015-06-29T09:24:02.144390 #2637:6a1ea4] INFO -- : MIQ(Host-Microsoft-verify_credentials): Connecting to WMI to verify credentials: [fqdn] -[domain\user] [----] W, [2015-06-29T09:24:02.248679 #2637:6a1ea4] WARN -- : MIQ(Host-Microsoft-verify_credentials): #<Errno::EPIPE: Broken pipe> [----] E, [2015-06-29T09:24:02.249536 #2637:6a1ea4] ERROR -- : MIQ(host_controller-update): Unexpected response returned from Cloud/Infrastructure Provider, see log for details
Just looking over miq-wmi-mswin.rb, my initial reaction is to remove the explicit call to WIN32OLE.ole_initialize. If you look at the win32ole.c source code you'll see this comment: "WIN32OLE.ole_initialize and WIN32OLE.ole_uninitialize are used in win32ole.rb to fix the issue bug #2618 (ruby-core:27634). You must not use these method." https://bugs.ruby-lang.org/issues/2618 I don't know if that will fix it right off but it's worth a shot.
Upon further review it looks like this is coming from the open3 call around the wmic call in miq-wmi-linux.rb, not the mswin version.
I see that Nick Carboni made a change related https://bugzilla.redhat.com/show_bug.cgi?id=1233815. In short, the code now uses awesome_spawn instead of open3. This also happens to solve this bug as well. https://github.com/ManageIQ/manageiq/commit/4ca5443b2a048df4e67efbc54d8816a3ee2861a3#diff-5649af8d36d8de353d3740927d454246 https://github.com/ManageIQ/manageiq/commit/d351e32c61468bee80435a9f6cd4ccd6d1df89e0#diff-5649af8d36d8de353d3740927d454246 Note that there is another potential issue with infrastructure hosts with domain names included in their primary DNS suffix, which may not resolve. This will cause failures.
I've tried with every combination of hostname, ipaddress with the correct domain\ user and it works every time. I'll keep an eye on it, and it is automated, so if there is an intermittent problem it should show up again. For now, verified on 5.5.0.8 https://10.16.7.25/vm_infra/explorer and several others.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2015:2551