[Solved (a hard way)] Update to 25.7.6 failed

[ednt]

Hi,

I did the trick with second search for updates after pkg was updated.
Update starts then like normal, but suddenly stucks and an Warning message appeard that the update failed.
Unfortunatley I don't know exactly where, but I think it was packet 4 of ..

Nothing works.

After a reboot I see the login prompt, but when I try to login, I get:

sh: /usr/local/libexec/opnsense-auth not found

and I'm not able to login.

Any idea?

[ednt]

Ok,

I booted into single user

Code Select Expand

used mount -u /
to make it writable

and created the file with

Code Select Expand

vi /usr/local/libexec/opnsense-auth
content:

Code Select Expand

#!/bin/sh

exit 0
Then

Code Select Expand

chmod 755 /usr/local/libexec/opnsense-auth
Now I can login after a normal boot.
But up to now no internet.

[franco]

The core package is missing and I suspect this is because the new package manager screws up by removing the core package early and then failing on some dependency before putting the core package back, see https://forum.opnsense.org/index.php?topic=49423.0


Cheers,
Franco

[ednt]

I used

Code Select Expand

ifconfig bce3 192.168.0.123 netmask 255.255.255.0
and

Code Select Expand

route add default 192.168.0.254
And I'm back on net.
/etc/resolve was still ok.
Now I started with

Code Select Expand

pkg update -f
and

Code Select Expand

pkg upgrade
This works now as expected. ( At the moment [122/127] Extracting ruby-3.3.9.1)

Next I will check if opnsense-update is available, then I will give it a try.

[franco]

> pkg upgrade

Do you have that particular output for me to prove the theory?


Cheers,
Franco

[ednt]

Hm...

Code Select Expand

opnsense-update
results in

Missing /usr/local/etc/pkg/repos/OPNsense.conf

I will try to create it by hand now.

[ednt]

Ups... there is no directory /usr/local/etc/pkg

Code Select Expand

mkdir /usr/local/etc/pkg

Code Select Expand

mkdir /usr/local/etc/pkg/repos

Code Select Expand

vi /usr/local/etc/pkg/repos/OPNsense.conf
with content:

OPNsense: {
  fingerprints: "/usr/local/etc/pkg/fingerprints/OPNsense",
  url: "https://pkg.opnsense.org/${ABI}/25.7/latest",
  signature_type: "fingerprints",
  priority: 11,
  enabled: yes
}

Now opnsense-update results in

Nothing to do.

How can I force it?

[ednt]

> pkg upgrade

Do you have that particular output for me to prove the theory?

Sorry, I'm at the console and can not copy anything.
But the FreeBSD upgrade worked without any problems.

[ednt]

Hm...

when I run

Code Select Expand

opnsense-update -gI get a log from the update before.
I attached it.

[ednt]

Can I run

Code Select Expand

opnsense-bootstrap.sh.in -r 25.7
Without loosing my (maybe) still available configuration of the interfaces and so on?

[franco]

opnsense-update needs an argument. opnsense-bootstrap can help but you're simply suffering from a partial upgrade it seems and steamrolling over it is not the best approach yet.

I suspect all you need to do is this if the opnsense repo configuration is there again.

# pkg install opnsense
# pkg upgrade


Cheers,
Franco

[franco]

Hm...

when I run

Code Select Expand

opnsense-update -gI get a log from the update before.
I attached it.

That's only the long from the previous update:

Currently running OPNsense 25.7.4 (amd64) at Wed Oct 15 07:50:31 CEST 2025
[...]
opnsense: 25.7.4 -> 25.7.5

and it went completely fine

[ednt]

As written: the log is not the last log, but was the output of opnsense-update -g

Code Select Expand

pkg install opnsense
Was Ok.

First run of

Code Select Expand

pkg upgrade
results in a segmentation fault.

A second try is now running. ( [99/104] Extracting ruby-3.3.9.1)

[ednt]

System is up and running again with original configuration!

Thank you Franco for your help.

But I think a wait an additional week, before I update the Master.
Maybe you will find something to prevent this behaviour.


Btw. I don't think that the before mentioned segmentation fault is comming from a RAM error.
The system is a HP Server DL360 with ECC RAM and the iLO Management of the Server shows no RAM error.
Also the tests at start are Ok.

[franco]

Question is when this segfault occurred, because I fixed one in pkg 2.3.1 that FreeBSD didn't fix and if you have the FreeBSD version that would explain a few things.


Cheers,
Franco