precise uec builds eat 100% cpu on launch, check-apt and rsyslogd

[Justin Randall]

Launching a UEC image (in my case on Eucalyptus in system mode), but I've seen similar reports in EC2. On launch, apt-check blocks initial login for MINUTES while collecting information for package updates. Afterward rsyslog eats 100% CPU failing to read proc/kmsg "Cannot read proc file system: 1 - Operation not permitted".

This effectively makes Ubuntu 12.04 precise images useless in the cloud and unsuitable as UEC sources.

I'll try to make a patch and provide fixed images, but this is appears to be a regression from issues reported from Lucid and really needs a proper test plan and fix to ensure new cloud images are usable.

[Ubuntu Foundations Team Bug Bot (crichton)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. It seems that your bug report is not filed about a specific source package though, rather it is just filed against Ubuntu in general. It is important that bug reports be filed about source packages so that people interested in the package can find the bugs about it. You can find some hints about determining what package your bug might be about at https://wiki.ubuntu.com/Bugs/FindRightPackage. You might also ask for help in the #ubuntu-bugs irc channel on Freenode.

To change the source package that this bug is filed about visit https://bugs.launchpad.net/ubuntu/+bug/1028682/+editstatus and add the package name in the text box next to the word Package.

[This is an automated message. I apologize if it reached you inappropriately; please just reply to this message indicating so.]

[Justin Randall (logicle)]

The initial hang for ssh logins may be due to /etc/update-motd.d trying to dynamically create a new motd and running updates-available. The UEC images assume there is a host resolvable called ubuntu-mirrir.localdomain, which is not always necessarily the case. Since these are typically short-lived ephemeral instances, reaching out to the Internet to retrieve package information as part of the motd creation process seems like a greater risk than benefit.

Adding "rsyslog stop" to rc.local might address the problems opening /proc/kmsg and spinning at 100% CPU.

[Justin Randall (logicle)]

I have verified that clearing out most of the dynamic content relying on Internet-facing connectivity from update-motd.d and adding /sbin/stop rsyslog to rc.local does solve the bug.

I suspect the problem with rsyslogd spinning at 100% cpu is a regression from Lucid, where kmsg is only accessible to root, but the process is running with reduced privileges and not detecting the problem.

# mount -o loop precise-server-cloudimg-amd64.img /mnt
# cd /mnt/etc
# vi rc.local

add /sbin/stop rsyslog to rc.local

# cd /mnt/etc/update-motd.d
# rm 90-updates-available
# rm 91-release-upgrade
# rm 98-reboot-required

[Ross Gammon (rosco2)]

Thanks for reporting this. It is quite an old bug now, and probably did not get any attention because it was not assigned to a package.
Having said that, I am not sure which is the right package to assign this bug to. Therefore I am assigning it to cloud-init, as I found a similar type of bug where this was done. If it is the wrong package, then hopefully someone looking at cloud-init bugs will have a better idea where to assign it so that it is followed up.

Also, as it is so old, perhaps the problem was fixed/went away in the meantime? Do you still see this problem?

[Launchpad Janitor (janitor)]

[Expired for cloud-init (Ubuntu) because there has been no activity for 60 days.]