System botched after upgrade attempt to 2.3.3
-
This just happened on my APU1, trying to upgrade from 2.3.2_1.
The first part of the upgrade seemed to go well (install new kernel), but after the reboot the unit is dead right after entering multiuser mode.
The console log shows:***** FILE SYSTEM MARKED CLEAN ***** Filesystems are clean, continuing... Mounting filesystems... Setting up memory disks...mdmfs: newfs exited with error code 255 mdmfs: newfs exited with error code 255 done. ___ ___/ f \ / p \___/ Sense \___/ \ \___/ Welcome to pfSense 2.3.2-RELEASE (Patch 1) on the 'pfSense' platform... Creating symlinks......done. Failed to write core file for process pkg (error 14) pid 134 (pkg), uid 0: exited on signal 11 Segmentation fault Failed to write core file for process pkg (error 14) pid 136 (pkg), uid 0: exited on signal 11 Segmentation fault Failed to write core file for process pkg (error 14) pid 137 (pkg), uid 0: exited on signal 11 Segmentation fault ERROR: It was not possible to identify which pfSense kernel is installed Failed to write core file for process php-cgi (error 14) pid 159 (php-cgi), uid 0: exited on signal 11 Failed to write core file for process php-cgi (error 14) pid 277 (php-cgi), uid 0: exited on signal 11 Failed to write core file for process php-fpm (error 14) pid 396 (php-fpm), uid 0: exited on signal 11 Segmentation fault fcgicli: Could not connect to server(/var/run/php-fpm.socket). Launching the init system...Failed to write core file for process php-cgi (erro) pid 401 (php-cgi), uid 0: exited on signal 11 Segmentation fault Starting CRON... Failed to write core file for process cron (error 14) pid 403 (cron), uid 0: exited on signal 11 Segmentation fault done. fcgicli: Could not connect to server(/var/run/php-fpm.socket). Failed to write core file for process php (error 14) pid 440 (php), uid 0: exited on signal 11 Segmentation fault Failed to write core file for process pkg (error 14) pid 445 (pkg), uid 0: exited on signal 11 Segmentation fault Failed to write core file for process pkg (error 14) pid 447 (pkg), uid 0: exited on signal 11 Segmentation fault Failed to write core file for process pkg (error 14) pid 448 (pkg), uid 0: exited on signal 11 Segmentation fault ERROR: It was not possible to identify which pfSense kernel is installed Failed to write core file for process php (error 14) pid 453 (php), uid 0: exited on signal 11 Segmentation fault pfSense (pfSense) 2.3.2-RELEASE (Patch 1) amd64 Tue Sep 27 12:13:07 CDT 2016 Bootup complete
Just about any process seems to segfault. The exact same thing happens when I boot with the old kernel.
I can get a prompt in single-user mode, but more segfaults:# vi upgrade_log.txt pid 25 (vi), uid 0: exited on signal 11 Segmentation fault
The upgrad log itself shows that it apparently started with the package and base system upgrades:
[84/88] Extracting pfSense-default-config-2.3.3: .... done [85/88] Upgrading pfSense-base from 2.3.2_1 to 2.3.3... [85/88] Extracting pfSense-base-2.3.3: ..... done ===> Keeping a copy of current version mtree ===> Removing schg flag from base files ===> Extracting new base tarball ERROR: It was not possible to identify which pfSense kernel is installed ...
Short of a clean re-install, is there anything I could try from the console, any way to roll back the botched update?
-
I just upgraded an APU with 4G nanoBSD on an SD card in it and that worked fine.
I guess APU1 is different to "vanilla APU"? -
I just upgraded an APU with 4G nanoBSD on an SD card in it and that worked fine.
I guess APU1 is different to "vanilla APU"?I mean the APU (PC Engines AMD G-T40E); I wrote "APU1" because there is now a newer version, the "APU2".
-
Personally, I'd start with a fresh full install image. Even if you were using NanoBSD, switch to a full install on that hardware.
-
Personally, I'd start with a fresh full install image. Even if you were using NanoBSD, switch to a full install on that hardware.
Yes, salvaging this hosed system is not worth it.
When the problems happened, I took another retired APU that still had 2.3.2 on it. The configuration was differernt, so I had to change the LAN IP address via a serial console and then restored the latest backup.
So, I was back in business soon enough, and the broken unit will now become the backup.
The troublesome update was already using a full install, I neglected to mention this.I'll try an update again with the replacement unit, but I'll use the console, so I should be better able to see what goes wrong, if it happens again.
-
After a clean install of 2.3.3 on my backup APU I made another attempt to upgrade the currently active unit from 2.3.2_1 to 2.3.3; this time I did it via the serial console, command "pfSense-upgrade".
Not surprisingly, it worked perfectly this time ;)