2

One of our boxes is running Ubuntu 8.04 LTS. Most Ubuntu package upgrades are trivial and don't affect availability much. How do I tell which packages are going to cause an outage before applying them?

2 Answers 2

6

Any package which installs/upgrades a running daemon will need a restart for that daemon. This is usually a very quick process, just a blip in service availability.

The only packages which need a reboot are kernel updates. You cannot upgrade a kernel without a reboot.

glibc updates may need you to restart a lot of services, or a reboot. However, given that most Unix services are still short lived due to the forking model, glibc updates are handled automatically.

The best way to know what is going to cause an outage is to actually apply them to a test machine and see what breaks. Testing is always a good idea.

2
  • 1
    Testing is, ultimately, the only way to know.
    – womble
    Jun 27, 2009 at 7:48
  • 1
    It's worth pointing out that the admin does not need to restart the services by hand, as the packaging will restart the required daemons automatically. As I've commented elsewhere, it's possible that the service will get stopped when you start the upgrade and only gets started again near the end of the process, so large upgrades risk having some noticable downtime for a particular service. Jun 27, 2009 at 8:35
0

Even the kernel upgrades that need a system restart will not cause you an outage - The old kernel will still be loaded and running until you restart the system in the future.

2
  • 3
    This is not entirely true. Many packages will incorrectly stop the service in the prerm script and start it again in the postinst. You will suffer outages for many upgrades. If you find any, please do file a bug. Jun 27, 2009 at 8:33
  • I was actually referring to kernel upgrades and system restarts. Editing my original answer...
    – Tzafrir
    Jun 27, 2009 at 17:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .