In last 3-4 days I've upgraded 10 servers (i386 and amd64) and couple of desktops from 6.06 and/or 7.10 to 8.04, something that makes system administrators sweat :). These were upgrades of not so important machines, but whole procedure was a test case for upgrades of important machines.
I taught there'll be plenty to write about, but all upgrades went without big or even medium sized problems. Biggest problem was waiting for all packages to configure :) For desktop upgrades, I've used 'update-manager -d', and for server upgrades all upgrading was done by do-release-upgrade program (from update-manager-core package). So, what could you expect during upgrade?
If you had NFS exports configured as:
/path 192.168.0.*(ro)
after upgrade, your clients won't be able to connect. I'm still not sure if that was correct way of specifying allowed clients, but it worked in 6.06. FWIW, now this works:
/path 192.168.0.0/255.255.255.0(ro)
That's it. That's the only thing that I had to fix after upgrades. Services upgraded on machines include: dovecot, postfix, nfs-kernel-server, samba, netatalk, chillispot, dhcp3-server, apache, squid, mdadm, hylafax, freeradius, rp-pppoe, fetchmail, sshd, no-ip, vsftpd, cupsd, openvpn, pptpd, mysql-server, ntop, ntpd...
Most of the custom scripts and oracle database 9.2 worked without any intervention. Very few scripts and configuration required adjustments.
If you have custom compiled applications (including PHP or Apache modules), don't forget to compile them again.
Before this week comes to an end, I'm planing to do a test upgrade of servers running redhat-cluster-suite (right, on Ubuntu!) and couple of mysql replications.
One can't say more than 'Great job everybody! Ubuntu server upgrade is easier than an upgrade of a Nokia mobile phone software :)'
Labels: ubuntu
