Thursday: I was going to take off about 1630 but ended up staying until 1830 mostly working on the zone file for smo. The thing that took such a long time was converting the saved secondary zone which is a semi-ordered not-quite-human-readable dump of data into an ordered file (mostly).
Gory details: We moved the primary publish point for the zone in preparation for shutting down smo name servers (but keeping smo domain names resolvable). It has a twisted thing that reads the hosts file and creates a dns zone from that — complete with CNAMEs where there were aliases for the machine, reverse entries, and it even makes TXT records for the comments. (We will be weaning people off of using /etc/hosts – on the production side we use DNS for just about everything anyway, save for “loghost” and a couple Sun machines that need /etc/hosts to learn their own names.)
Friday: Managed to leave work about on time, just some last-minute sending of announcement mail, disclosing the pre-pushing we were doing in case anyone reboots their machine early. Once home, there was lots of cleaning to be done because we were expecting company (M’s parents came up Saturday morning and stayed until Sunday evening.) I got to sleep around 0000 I think.
Saturday: Up at 0600 to meet BW at the data center at 0700. I arrived at 0656 with Krispy Kreme in hand. At Albertson’s I saw something which disturbed me: “Self Checkout Stations” (four). Apparently you can scan it, bag it, slide your card or insert cash, and not have to talk to a human unless something goes wrong. Scary. (Yes Albertson’s by my house sells Krispy Kreme). We then proceeded to shut down user workstations, shut down servers, bring back up servers, drive to Palo Alto, and work on user workstations.
More gory details: Workstations mostly shut down to single user mode just fine (about 1 hour). Servers shut down fine. Reconfiguring NIS and autohome on the main home server took about 30 min. Bringing servers back up was kind of a confusing collection of 1. learning which servers depend on which other servers for proper booting and proper operation, and 2. figuring out which ones had serial console and which had vga only. Bringing back the remaining servers to proper operation took from like 1000 to like 1300 (ouch- we were hoping to leave for PA about 1130).
After a stop at Carl’s Jr. at company expense, we headed for PA. I spend about the first hour trying to get my own workstation and LP’s (my neighbor’s) to come up cleanly- for the most part it didn’t log me in or didn’t start X properly, indicating no home directory (even though my home directory was supposed to be local on my box). Had to hit reset a lot, because I don’t have sudo access, and couldn’t read the root password correctly off the page, so any root operations had to be done by way of reset, ^C, >>>boot -fl s, mount -u /, mount /usr, setenv TERM vt100, type commands or edit files, then shutdown -r now to see if it works. After being unsuccessful at this for the first 45 min I decided to try someone else’s workstation that wasn’t Ops special setup, after 15 minutes of that it was more of the same. BW came to find me and indicated that he was having similar luck with the first 3 servers, due to NIS not working at all really, and that he had called Compaq field service for help.
The solution (after another hour) turned out to be that we had to re-touch and re-make the NIS settings once the slaves were up. (Note to self: next time make the changes to the master and push them out before taking other slave servers down. Additional note to self: NIS is brain-dead and stupid and remind me to get rid of it Real Soon Now — my preference would be to have both NIS and NFS gone in 6 weeks.) Anyway after repushing to the 3 slaves, the clients seemed to be fine.
Then comes the tedious part. From about 1600 to about 2100 we were reconfiguring desktops. Each desktop needed edits to: resolv.conf, rc.config (ifcfg-eth0), routes (network), hosts, ntp.conf. Only rc.config (most) and ifcfg-eth0 (about a third) had been pushed so on most workstations I was editing 5 files. I got through about 30 workstations in about 4-5 hours, and spent 2100-2130 at typing up a quick list of done vs. not done for BW to finish the next day.
Got home and watched some TV with M and her parents (stopping for Taco Hell and 7-11). Got to bed about midnight again.
Sunday: I had told BW that I would rather stay later on Saturday than come in Sunday (which I had done) so I didn’t go in to work. Still managed to get up about 0840 due to waking up M and M waking me in turn — for some reason I couldn’t fall back to sleep without snoring pretty loud, so we both just got up.
Went with M’s parents to Original Pancake House. Everything was great there as always… the only downside being that they didn’t have non-dairy creamer (and like 2 busgirls, 1 server and 1 hostess sort of acted like they didn’t know what we were asking). After that, M’s dad helped me to take the tank off the toilet and replace the guts. We got the right part on the 2nd trip to the hardware and got the toilet tank put back on with no leaks on the 2nd try. Seems to work fine! Watched some more TV, then went to dinner with misty_shadows and traveller_blues also in tow. Said good-bye to M’s folks, sat and chatted with m_s and t_b for a bit more.
That is all.