Difference between revisions of "Upgrade Lenny"

From Roaring Penguin
Jump to: navigation, search
Line 53: Line 53:
(a) and then (1) on all machines, then (a) and (2) on all machines, and
(a) and then (1) on all machines, then (a) and (2) on all machines, and
so on.
so on.

Revision as of 15:43, 29 January 2014

Upgrade Debian Lenny to Squeeze

You must go from Debian release to Debian release, unlike CanIt itself which allows "jumping" over interim releases.

Since Lenny is no longer supported, we recommend upgrading the Debian version to Squeeze (6.0). (DO NOT upgrade to Wheezy (7.0) without first contacting us.) Here are the commands you should run as root. We recommend doing the upgrade during Roaring Penguin's office hours (Mon-Fri, 9:00-17:00 EST) in case you run into trouble.

1) Make sure your existing CanIt installation is up-to-date:


2) Now upgrade the OS to Squeeze:

   upgrade-canit-appliance squeeze

Then you'll see messages scroll by, as the machine downloads, unpacks, and re-starts assorted services. (This all only delays mail during the re-start phase).

3) Since Debian has changed the way it distributes firmware, you may need additional packages. Even if you don't need them, installing them won't hurt so we recommend running:

    apt-get install firmware-linux-free firmware-linux-nonfree firmware-bnx2 firmware-bnx2x

(That's all one long line in case it got line-wrapped by a mail reader.)

If, after running this command, you still receive warnings about "Possible missing firmware", please reply with the exact error message. (lsmod and check for unusual modules just in case.)

4) Reboot the machine

    (tail -f /var/log/mail.log or wherever they are to see if working)

NOTE: The above commands are for a single machine. If you are upgrading a cluster, then you should modify the steps as follows, starting with the DataBase machine:

a) Before every command above, run:

    /etc/init.d/canit-system stop-gracefully

b) After the command in step (4), run:

    /etc/init.d/canit-system start

c) Also, all commands should be run in parallel. That is, run (a) and then (1) on all machines, then (a) and (2) on all machines, and so on.