UEC Node controller loses connectivity

[Russell de Moose]

Natty 11.04, UEC (1 front-end, 2 NC), Eucalyptus 2.0, Lucid 10.04 image for instances

I had a working UEC environment, where the euca-describe-availability-zones verbose would show the resources of the 2 combined NCs. I have 2 instances running on each NC. Yesterday I could not access the instances from the 2nd NC, and the CC could not communicate with the NC. They did not show up after running euca-describe-instances, but the NC DID show up when running euca_conf --list-nodes, and checking the nodes.list file in /var/lib/eucalyptus. Several hours later, the same commands (euca-describe-availability-zones verbose and euca-describe-instances) show the previously unavailable resources, and now the instances are again accessible. It seems as if the NC simply disappeared for several hours, in which time I could still SSH into the NC, CC, and verify that all services were running.

ProblemType: Bug
DistroRelease: Ubuntu 11.04
Package: eucalyptus-common 2.0.1+bzr1256-0ubuntu4.1
ProcVersionSignature: Ubuntu 2.6.38-10.46-server 2.6.38.7
Uname: Linux 2.6.38-10-server x86_64
.etc.eucalyptus.eucalyptus.cc.conf: CC_NAME="webness"
Architecture: amd64
Date: Thu Aug 4 12:05:14 2011
InstallationMedia: Ubuntu-Server 11.04 "Natty Narwhal" - Release amd64 (20110422)
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: eucalyptus
UpgradeStatus: No upgrade log present (probably fresh install)