** Description changed:
- I spawned a lucid AWS instance and used some cloud-init config to
- perform some tasks for me.
+ === Begin SRU Information ===
+ [Impact]
+ * If a user launches a 10.04 cloud-image with a user-data script, and then
+ does a upgrade to 12.04 and 'reboot's to complete the upgrade, the
+ user-data script will incorrectly execute again on boot.
+
+ This is because the cloud config module that executed user-data changed
+ names between 10.04 and 12.04 from 'user-scripts' to 'config-scripts-user',
+ but nothing was in place to adjust the "already ran" marker file
+
+ [Test Case]
+ * launch an instance with some user-data:
+ $ cat lp-1049146-user-data.txt
+ #!/bin/sh
+ echo "====== $(date -R) [$0] =====" | tee -a /root/user-data.log
+
+ $ ubuntu-ec2-run --key=your-keyname --instance-type=m1.medium \
+ --user-data-file lp-1049146-user-data.txt lucid daily
+ # us-east-1/instance-store/ubuntu-lucid-daily-amd64-server-20121108
+
+ * upgrade the image and then do-release-upgrade it, then reboot
+ $ IP=THE_IP_ADDRESS
+ $ ssh ubuntu@$IP
+ % sudo apt-get update -q && sudo apt-get upgrade -y
+ % do-release-upgrade --frontend=DistUpgradeViewNonInteractive
+
+ # you could install the patched new version of cloud-init by hand here.
+ % sudo reboot
+
+ * see that the user-data has run more than once
+ # upon reboot, you'll see that the user-data script ran again (twice)
+ $ sudo cat /root/user-data.log
+ ====== Thu, 08 Nov 2012 17:07:52 +0000
+ [/var/lib/cloud/data/scripts/part-000] =====
+ ====== Thu, 08 Nov 2012 17:26:50 +0000
+ [/var/lib/cloud/instance/scripts/part-000] =====
+ ====== Thu, 08 Nov 2012 17:26:50 +0000
+ [/var/lib/cloud/instance/scripts/part-001] =====
+
+ [Regression Potential]
+ * There is little potential to make anything worse here. The relevant
+ change is really only to touch a file in the post install that will
+ serve as a marker for the newer version of cloud-init, so it realize
+ that the user-data has already been executed once.
+
+ === End SRU Information ===
+
+
+
+ I spawned a lucid AWS instance and used some cloud-init config to perform
some tasks for me.
All is good.
Then, as part of the task I was performing, I upgraded that lucid
instance to precise using do-release-upgrade. It failed, but unrelated
to cloud-init (a postinst from a package failed).
I then rebooted the instance. When it came back up, the cloud-init
config I supplied the first time was run again, as if this was the first
time the instance booted.
smoser had access to before and after reboot, and asked to add this to the
bug:
# for x in /var/lib/cloud/instance/sem/*; do [ -f
var-lib-cloud/instance/sem/${x##*/} ] || echo ${x##*/}; done
config-byobu
config-ca-certs
config-chef
config-grub-dpkg
config-keys-to-console
config-landscape
config-mcollective
config-phone-home
config-rightscale_userdata
config-rsyslog
config-salt-minion
config-scripts-per-instance
config-scripts-user
config-set-passwords
config-ssh-import-id
config-timezone
$ ls var-lib-cloud/instance/sem/
config-apt-pipelining config-puppet config-updates-check
config-apt-update-upgrade config-runcmd consume_userdata
config-locale config-set_hostname user-scripts
config-mounts config-ssh
--
You received this bug notification because you are a member of Ubuntu
Server Team, which is subscribed to cloud-init in Ubuntu.
https://bugs.launchpad.net/bugs/1049146
Title:
cloud-init runs again after reboot in release upgraded instance
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1049146/+subscriptions
--
Ubuntu-server-bugs mailing list
Ubuntu-server-bugs@lists.ubuntu.com
Modify settings or unsubscribe at:
https://lists.ubuntu.com/mailman/listinfo/ubuntu-server-bugs
