On Thursday 17 February 2011 17:33:24 Mark wrote: > Realizing this is dependent on computer specs, just curious what some of > the people on this list have experienced for how much time it took to do > the Lenny to Squeeze upgrade, (assuming a fully up-to-date Lenny system). I took a few hours for me to take backups, edit APT configuration, perform the partial upgrade, correct configs, preform the full upgrade, correct configs, reboot, and correct configs. Then later, I had to correct more configs based on weird behaviors I was seeing on some of the public facing services. I'd say give yourself 4-8 hours follow the release notes though the upgrade. You can upgrade multiple systems at once. Most of this time is spent waiting, so you can do multiple systems at once. There will be occasional debconf prompts, so it's difficult to completely automate and you'll need to keep the systems you are working on straight. Depending on what services you host and how much you've changed their configuration files, you'll want to allocate 1 extra hour at upgrade time and 1 hour during the following 5-7 days, per service. For me, ClamAV was the easiest, it was able to migrate to the new version without any interactivity. Postgresql and SpamAssassin come in second. I didn't have to twiddle Postgres configurations but I did have to "manually" do the cluster upgrade. (I didn't have to do anything more complex than use the pg_*cluster commands.) Spamassassin's configuration file needed a manual merge because I have custom rules and weights, but I also wanted to use the new, optional short-cutting. Drupal 6 was pretty easy, but I accidentally locked myself out for a while with the site in maintenance mode. :( Dovecot was a real problem, with cmusieve being dropped in favor of sieve and having slightly different configuration options on top of having a fairly customized Dovecot configuration that ucf couldn't handle. Even manually merging, I had to fix it up a half-dozen times.[1] Exim cleanly upgraded, but did have a unmergable configuration file -- with a little research I determined I wasn't actually using that file so I could use the package maintainer's version without worry. Approx moved from running under /etc/init.d to needing an inetd, but it is a pretty simple service so it wasn't much trouble. The leftover init script was preventing a migration to dependency-based startup, but insserv has good instructions for that. All considered it is pretty easy, but do make sure an allocate some time to remediation for about a week after the upgrade if you or anyone else depends on services. [1] I might still not have it quite right -- it seems that deleted messages sometimes reappear, but I'm not sure if that is a bad client or some bad imapd settings in dovecot. -- Boyd Stephen Smith Jr. ,= ,-_-. =. bss@iguanasuicide.net ((_/)o o(\_)) ICQ: 514984 YM/AIM: DaTwinkDaddy `-'(. .)`-' http://iguanasuicide.net/ \_/
Attachment:
signature.asc
Description: This is a digitally signed message part.