Update 2100 to 23.01 advice
-
Apologies for newbie questions. I have a Netgate 2100 running 22.05 with Avahi 2.2_1 and pfBlockerNG-devel 3.1.0_11. I've been following the upgrade discussions here and want to try to avoid problems. Should I first delete Avahi and pfBlockerNG-devel prior to upgrading? pfBlockerNG-devel is set to Keep Settings on upgrade. Will the settings carry across to the new (non-devel) pfBlockerNG? I don't see a Keep Settings for Avahi but maybe it's simple enough it's not needed for this package?
-
@terryzb In general:
I haven't used Avahi, no comment there.
I do always uninstall pfBlocker and reinstall after.
pfB-devel was migrated to pfB(non-devel) for 23.01 so you can install the non-devel version.
When you start the upgrade let it finish...it will likely take at least 10-15 minutes if not more. If you reboot early it will be corrupted/partial install.
The upgrade will reinstall any packages you don't uninstall.There is a bug on 1100/2100: https://redmine.pfsense.org/issues/13967. I know some with a 2100 have posted they updated successfully. I have not tried any yet, I usually like to watch for bugs for at least a few days. If you do run into trouble you can get the image file from Netgate to reinstall, no support contract necessary.
-
@steveits Thank you for pointing out the potential bug on 1100/2100s. Good advice to wait a few more days at least.
-
Yup we are still trying to narrow down the exact circumstances that triggers that bug. So far I've been unable to replicate it....
-
@stephenw10 Try and see if it’s connected with making a pre-upgrade reboot or not. It could be something that only happens on boxes with a fair amount of uptime.
-
It's almost certainly because the original filesystem layout used for the 1100 and 2100 had a very small EFI partition and the upgrade code is trying to write into it the current EFI loader which is too large. However no combination of installs and upgrades I have tried have hit it. We are still digging into it here.
However if you have the original layout in which the EFI partition is 800K you are more likely to hit it IMO.
[23.01-RELEASE][root@1100-3.stevew.lan]/boot/efi: gpart show => 1 15273599 diskid/DISK-2EC51400 MBR (7.3G) 1 1600 1 efi (800K) 1601 70012 2 fat32 (34M) 71613 15201987 3 freebsd (7.2G)
Steve
-
I received my 2100 last year and I see 200MB for the efi, which sounds like good news.
[22.05-RELEASE][root@fw.home.arpa]/boot/efi: gpart show => 1 15273599 mmcsd0 MBR (7.3G) 1 409600 1 efi (200M) 409601 70012 2 fat32 (34M) 479613 14793987 3 freebsd [active] (7.1G) => 0 14793987 mmcsd0s3 BSD (7.1G) 0 16 - free - (8.0K) 16 14793971 1 freebsd-zfs (7.1G)
-
@stephenw10 I have a 2100 at home with the small partition and it was an early one. I can reinstall if needed (which would convert to ZFS) but if I just try it for you is there anything special to watch for besides the upgrade log, or look at beforehand?
-
Just note what version it's upgrading from and the filesystem type. The complete upgrade log might show something if it does fail.
Still been unable to trigger it even doing something extreme like upgrading from 2.4.5 dircetly. Which I'm sure the people who are hitting this are not.Steve
-
-
-
@stephenw10
So, I did have this problem. :( Ufs, small efi partition, error on the EFI loader. An early 2100, Serial: 19302xxxx.Recovery was OK, though longer than expected because Etcher apparently will warn you the drive is too small if it's a 256 MB USB, but not any other combination I tried ٩(͡๏̯͡๏)۶ until I started reading the sizes. That'll teach me a lesson to actually write the USB stick first. Didn't realize 23.01 was that much larger. I installed 22.05 because it would fit on my 4 GB USB, restored my config, and upgraded to 23.01.
I have the config if you want it though I would be surprised it's config related. I did have RAM disks for /tmp (512M) and /var (1G), both mostly empty, I uninstalled pfBlockerNG-devel, and left apcupsd installed as the only package (with the UPS connected).
I also have the full log shown in the GUI during the install, and the 2100 s/n and ID, or whatever else you need.
I have the console output after too which (after the 'usbrecovery', after the power off, on 22.05) had a couple of things that caught my eye:
failed to open /dev/mmcsd0: Operation not permitted
Growing root partition to fill device
...it went on to resize mmcsd0s3 and "Filesystems are clean" so I'm assuming that's normal.
...followed by four "1076809728:error:2506406A:DSO support routines:dlfcn_bind_func:could not bind to the requested symbol" errors.The restore on 22.05 did say it was reinstalling packages, and the dashboard warning went away upon refresh. I was annoyed enough at myself (plus cat on keyboard and entire family now in the room with me) I didn't double check both before upgrading again, or uninstall pfBlocker again, but after 23.01, apcupsd and pfBlocker were not installed, so that's interesting.
I see I got an email at the time that "Package reinstall process was ABORTED due to lack of internet connectivity"...which is interesting since I got the email and Internet seemed just fine.(not really trying to hijack this thread...sorry OP)
-
Can you upload those here?
https://nc.netgate.com/nextcloud/s/Yrt4SBdMcNpYkRS -
@stephenw10 uploaded