Three days ago I lauched an EC2 Ubuntu Precise micro instance on Amazon Web Service. AMI: ubuntu/images/ebs/ubuntu-precise-12.04-i386-server-20120424 (ami-e7e8d393)
This is my very first server experience so I'm going slow and reading and observing the system. During this three days I didn't install any new packages other than appyling four batches of updates. Some of them were security updates, some were not; some were kernel updates, some were not. My var/log/apt/history.log is attached.
As a result:
# uname -a
Linux [some-internal-IP] 3.2.0-24-virtual #38-Ubuntu SMP Tue May 1 17:00:51 UTC 2012 i686 i686 i386 GNU/Linux
If it might be useful:
# ls -l /etc/resolv.conf
-r--r--r-- 1 root root 91 May 23 09:59 /etc/resolv.conf
# cat /etc/resolv.conf
domain eu-west-1.compute.internal
search eu-west-1.compute.internal
nameserver 172.16.0.23
I didn't do any configuration to the default system of any kind by myself. Up till now I just play with the basic commands such as whoami, top, ls, cd etc. and to observe the update/upgrade processes.
As you can see from the attached file, my fourh and last upgrade is from resolvconf 1.63ubuntu11 to resolvconf 1.63ubuntu14 and the upgrade resulted with the error: "resolvconf: Error: /etc/resolv.conf isn't a symlink, not doing anything." What I get from connection up to that error is attached.
I searched some on the Web and figured out that issues having something to do with the error "/etc/resolv.conf isn't a symlink, not doing anything" is being experienced for Ubuntu for years. The causes for all of them might not be the same but seems thet have to have something in common. I'm not an expert on such things. Some places on the Web for such discussions: http://ubuntuforums.org/showthread.php?t=1193585 , http://askubuntu.com/questions/137037/network-manager-not-populating-resolv-conf , http://eric-b.net/?p=71 . There are discussions on Lauchpad as well: Bug #922677 is an example. Bug #324233 (with 3 duplicates) sounds more alarming with what Thomas Hood (jdthood) wrote on 2011-12-03 at the end of his long comment with a patch: "Who am I? I am one of the maintainers of resolvconf in Debian. I would prefer to leave this work up to Ubuntu folks, but these issues have been left unresolved for years. Something has to be done."
On a server installation such as EC2, with no extra package installed other than offical updates, with no personal configuration -all being the defaults- a bug known for years, for such a critical component for a server -again- surfaces up. Does it make sense?
I'm not adding any more files, not adding the files the original reporter had attched as well because I don't know what you need. If you'll ask me for more files outputs, I can send them.
I don't suppose so but if this a very interesting, atypical situation calling for deep investigation, I can give the control of the EC2 instance for some time to a friend with reputation dealing with such issues. The instance is free of charge for 1 year and includes nothing except its default being. I can terminate it and launch a new one any time.
Until I hear something from you here, I won't do anything, any update or so to that system for 2-3 days. ıf I can not hear something soon maybe I'll terminate it and can't answer probable questions anymore.
Regards.
PS: On the system in question, I can already view Web pages via w3m, if it means something in some way.
It's not a bug, resolvconf is just letting you know that it won't update the file as it's not a symlink, nothing wrong there.
I might agree that showing "Error" might be a bit excessive and that we might want to replace that by "Warning".
The problem, if we can consider it to be one is that one of the cloud image tools is replacing the /etc/resolv.conf symlink by a static file instead of updating /etc/network/