F12 on boot brings up boot selection menu. Choose boot from onboard NIC. PXE works. Doing nothing tries to install windows, so hit F8. Chose to configure static IP even though DHCP worked. Otherwise debathena installer runs into an error while configuring networking. 1st attempt: Choose "normal Debathena install" (not cluster). Manual partitioning. Entire disk, one large LVM container, containing swap and root partition. (Note well, this is a little bit adventurous: the "old" way was to create a /boot partition outside of LVM. However newer versions of GRUB can read a /boot inside LVM logical volume, so there is no longer a need for a separate boot partition.) root filesystem ext4, added "noatime" mount flag because it is SSD. grub-install failed: "Embedding is not possible, but this is required for RAID and LVM install." Gave up. 2nd attempt: Same as before, but no LVM. one partition for root filesystem, another partition for swap. partitioning tool complains about lacking a bios-grub partition. go back and create a 1 MB bios-grub partition at the beginning of the disk, followed by the root filesystem partition, followed by swap. Chose Debathena flavor "login-graphical". This succeeded to the end of the installer, but decided to try again, because we want "workstation" flavor not "login-graphical" after reading further documentation. 3rd attempt: Try LVM again, this time with bios-grub partition ahead of the single giant LVM partition. Inside it swap (24 GB), and root (the remainder of the space). Ext4, noatime, same as before. Local user account "sipbadmin". Declined to enable automatic updates, because we assume Debathena configuration later will do this. Did not install any additional software, because we assume Debathena configuration later will do this. This time grub-install worked fine. Chose Debathena flavor "workstation". Install completed, reboot, failed to find bootable disk. Went to BIOS setup and changed boot preference to Legacy instead of UEFI. (This was probably the original reason it failed to boot, triggering this adventure of reinstalling.) Upon further research, changed noatime to relatime in /etc/fstab because "noatime" supposedly causes problems with mutt, and people reportedly use mutt around here. Added rounds=300000 to /etc/pam.d/common-password so password hashing takes about 0.25 seconds. sipbadmin password and root password are office root password. restored ssh_host_* from old machine. restored /etc/krb5.keytab from old machine sudo su kinit kenta aklog sipb cd /afs/sipb/machine/office/bin ./syncsipb installed mosh ---- updates 2016-08-05 /etc/network/interfaces did not have a em1 interface; it only had a loopback lo interface. (This caused, somehow, NetworkManager to attempt to manage the interface, despite /etc/NetworkManager/NetworkManager.conf [ifupdown] managed=false , creating /etc/NetworkManager/system-connections/Wired\ Connection\ 1 ) This caused the side effect of the system not having a default search domain MIT.EDU, so hosts on MIT have to have their full hostnames spelled out, e.g., cvp.mit.edu instead of cvp. Adding the em1 interface stanza fixed the DNS search domain issue. It does not fix the issue of logging in causing an immediate log out after a few seconds. problem spotted in syslog Aug 5 22:13:18 JEBEDIAH-KERMAN gnome-session[3497]: WARNING: Application 'compiz.desktop' killed by signal 11 Aug 5 22:13:18 JEBEDIAH-KERMAN gnome-session[3497]: WARNING: App 'compiz.desktop' respawning too quickly Aug 5 22:13:18 JEBEDIAH-KERMAN gnome-session[3497]: CRITICAL: We failed, but the fail whale is dead. Sorry.... possible bug https://bugs.launchpad.net/ubuntu/+source/fglrx-installer/+bug/1072329 workaround worked for me https://bugs.launchpad.net/ubuntu/+source/fglrx-installer/+bug/1072329/comments/6 trying logging in again, cannot log in. potentially useful message in ~/.xsession-errors: init: gpg-agent post-stop process (20188) killed by TERM signal init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd main process ended, respawning init: at-spi2-registryd respawning too fast, stopped there is a crashlog for compiz in /var/crash -- This is weird: su Password: root@JEBEDIAH-KERMAN:/var/crash# aptitude install -R apport-retrace The following NEW packages will be installed: apport-retrace python-apport{a} python-problem-report{a} 0 packages upgraded, 3 newly installed, 0 to remove and 0 not upgraded. Need to get 97.2 kB of archives. After unpacking 908 kB will be used. Do you want to continue? [Y/n/?] Get: 1 http://mirrors.mit.edu/ubuntu/ trusty-updates/main python-problem-report all 2.14.1-0ubuntu3.21 [9,634 B] Get: 2 http://mirrors.mit.edu/ubuntu/ trusty-updates/main python-apport all 2.14.1-0ubuntu3.21 [75.2 kB] Get: 3 http://mirrors.mit.edu/ubuntu/ trusty-updates/main apport-retrace all 2.14.1-0ubuntu3.21 [12.3 kB] Fetched 97.2 kB in 0s (911 kB/s) debconf: Perl may be unconfigured (Can't locate strict.pm: Permission denied at (eval 1) line 2. BEGIN failed--compilation aborted at (eval 1) line 2. ) -- aborting Selecting previously unselected package python-problem-report. (Reading database ... 423879 files and directories currently installed.) Preparing to unpack .../python-problem-report_2.14.1-0ubuntu3.21_all.deb ... Unpacking python-problem-report (2.14.1-0ubuntu3.21) ... Selecting previously unselected package python-apport. Preparing to unpack .../python-apport_2.14.1-0ubuntu3.21_all.deb ... Unpacking python-apport (2.14.1-0ubuntu3.21) ... Selecting previously unselected package apport-retrace. Preparing to unpack .../apport-retrace_2.14.1-0ubuntu3.21_all.deb ... Unpacking apport-retrace (2.14.1-0ubuntu3.21) ... Processing triggers for man-db (2.6.7.1-1ubuntu1) ... Setting up python-problem-report (2.14.1-0ubuntu3.21) ... Setting up python-apport (2.14.1-0ubuntu3.21) ... Setting up apport-retrace (2.14.1-0ubuntu3.21) ... === there are some crash logs in /afs/sipb/user/kenta/reinstalling-jebediah-kerman-crashlogs _usr_bin_compiz.18556.crash.xz is a regular crash log, athena user kenta _usr_bin_compiz.1000.crash.xz demonstrates crash with a local user, and with the OpenAFS PPA not enabled. === nth attempt command line install of debathena (as opposed to pxe) fails the same way. -- since we are doing vanilla ubuntu, why not 16.04? unable to get jeb or otis-oracle to boot off of either of sipb's usb sticks, after writing 16.04.1 amd64 server image with mkusb. weird. try burning cd. -- first attempt at burning a cd failed. using gui second attempt, mini.iso in the netboot directory, used wodim and coasterless site. works. need to do expert install to avoid ipv6 autoconfiguration which causes a hang when connecting to security.ubuntu.com still need to add stanza to /etc/network/interfaces the interface is now named eno1 --