Updating to current from r107

Getting Started with IPFire
1

I’ve been tasked with upgrading our out of date ipfire instance. I had started drafting a forum post asking if it would be better to attempt a series of sequential updates from 107 to latest (121 was offered), or to backup / fresh install / restore. However, due to a browser drawing issue I accidentally clicked the update button and we’re now at 141 and awaiting a reboot.

When I misclicked yesterday, I saw the pakfire console stepping through updates from 107 to 121. The connection timed out a couple time, but I got reconnected and was placed back into the console. When it got to release 121 I was dropped back into the pakfire gui, which showed an available upgrade from 121 to 131.

Today when I looked at the logs, I discovered that it continued upgrading, silently in the background.

15:38:19 pakfire:  CORE UPGR: Upgrading from release 107 to 121
...
15:46:32 pakfire:  PAKFIRE UPGR: core-upgrade-121: Finished.
15:46:39 pakfire:  PAKFIRE INFO: Pakfire has finished. Closing.
15:46:39 pakfire:  PAKFIRE INFO: IPFire Pakfire 2.21-x86_64 started!
15:46:39 pakfire:  CORE UPGR: Upgrading from release 121 to 131
...
15:54:30 pakfire:  PAKFIRE UPGR: core-upgrade-131: Finished.
15:54:30 pakfire:  PAKFIRE INFO: Pakfire has finished. Closing.
15:54:31 pakfire:  PAKFIRE INFO: IPFire Pakfire 2.23-x86_64 started!
15:54:31 pakfire:  CORE UPGR: Upgrading from release 131 to 141
...
16:01:23 pakfire:  PAKFIRE UPGR: core-upgrade-141: Finished.

At this point it started logging a mirror error, which is probably why it didn’t continue to the latest release of 144.

Also, the pakfire gui now shows inaccurate list update times:

Last update made 20h 26m 36s ago.
Last server list update made 18402d 17h 27m 59s ago.
Last core list update made 18402d 17h 27m 59s ago.
Last packages list update made 18402d 17h 27m 59s ago.

Questions:

  1. Is this background updating a known bug? (I’d guess so.) Is it fixed?
  2. What’s the risk from delaying the requisite reboot until our regular maintenance window two weeks from now?
  3. Is the mirror error a known issue that will be resolved with the reboot?
  4. Are the strange list update times another thing likely to be resolved after reboot?
  5. Even though I know 144 is the latest release, No update to 141 is offered. Is this due to the mirror error, or is there something else going on?
  6. If we ever get this far behind again, is it better to go through the sequential updates, backup / fresh install / restore, or something else?

Thanks!

2

Hi @mathuaerknedam, you are wellcome. .

An advice. Now where you are, make a backup of the configuration and do a clean installation.

Another option, if you do not want to touch the hard disk, change it for another one, in that other one you install the clean 144 and the hard disk that you have removed, you install it via USB and you mount and manually copy the configuration files. (var / ipfire)

This second option I have done on several occasions and thus you will avoid headaches

You tell us.

Regards.

1 Like
3

If the hardware allows it, go x64.

1 Like
4

Changed partitioning scheme and sizes have been introduced since core 107. It is possible that your update ran out of space and stopped at core 141.

You should get much better outcome by downloading your backup file for core 141, reinstalling at core 141 and then the update to core 144 is likely to proceed. Core 141 is available from some mirrors.

I caution against reinstalling core 144 directly, because some of the settings in your backup would be incompatible.

Why did you get so far behind ? Each update takes only a few minutes. I have done some at CLI and it looks feasible to do as a chron job, during silent hours, if you must.

5

cronjobs are currently unsupported.

6

Thanks for all the advice.

We’re currently on x86_64, and we’ll certainly continue that.

the closest any partition (besides /boot) is to being full is / with 486m free. /var has 224G free, and no partition is anywhere close to running out of inodes.

I’m not sure why it hasn’t been updated in so long, but I don’t expect it to be a problem going forward.

7

Is your installation virtual or baremetal?

8

With core141 we ship a new pakfire version and we have decided not automatic restart it. (this was the reason for the background updates at earlier version changes)
At core141 the dns cache needs a reboot on some connection configurations to get the isp dns server.
Without working dns pakfire cannot download new lists or reach the mirrors.

9

@pike_it Bare metal.

@arne_f Thanks for the explanations.

10

It works even on bare metal, but it’s quite longer.
You can try to

  • clone your firewall disk into another disk or image (clonezilla?)
  • resize the partition on the disk, bringing a bit more “breath” to your /boot mountpoint (512MB? 1GB? I don’t know how big is now) with a live disk (i generally use Lubuntu if the appliance has VGA port)
  • reebot into ipfire, checking the size of the partition and the free space
  • update

In virtual environment is quite faster (cloning drive is just a matter of copied files/directories with a shutted down guest) and without clonezilla intervention.
Shutting down the firewall maybe can be useful also to wipe some dust out :wink: