Menu

Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Messages - mooh

#1
If your HA and Dirigera are part of the same fabric, you are using a multi-admin setup. In that case both hubs need to be on the same (V)LAN segment.

As I tried to explain, either keep them on the same (V)LAN in multi-admin mode or remove the devices from HA, add the Dirigera as a Matter hub to HA. This will expose devices connected to the Dirigera via Matter to the HA. Then, you may move the HA to another network because HA will not try to talk to the devices via their ULAs directly.
#2
It looks like the sequence numbers of cloned rules are identical to that of the original rule. I understand that this is technically correct and it didn't cause problems for me, yet. But what other feature is there that keeps the PF evaluation order the same as the rules display in the UI? I would expect the UI to recalculate the sequence to match the place in the UI list, the same way as the sequence is recalculated when rules are moved around.
#3
Quote from: mooh on November 16, 2026, 02:16:27 PMIt's a bit unclear to me what you are trying to achieve. Matter is an application protocol. Matter devices are registered to a Matter controller. That's called a "fabric". The Matter controller usually happens to live on a border router. Your Dirigera hub is a Matter controller and border router that connects devices on Ethernet and Thread networks. So, if you want to communicate on the Matter level, connect to the controller via its name or IP, like you do with any other device on your local networks.

If however, you want to communicate to your Matter devices bypassing the controller, you will need to explicitly configure routes. But, as you already found out, you have no control over the ULAs. This, in fact, is by design. Thread is a radio network. If communication between devices on that network breaks down for whatever reason, the network will reorganise into so called partitions, i.e. partial networks where communication still works. The partitions will reconfigure and choose a new ULA. Same thing happens when partitions eventually rejoin.

To work around this simply add the Dirigera controller as a Matter hub to your Home Assistant, if there is support for this. That way, HA and Dirigera will talk Matter protocol. Don't communicate with your Matter devices via low-level IP, i.e. don't use multi-admin mode. It only drains batteries faster.

If that's not an option, follow Patrick's advice and put your HA into the same network as the Dirigera hub (possibly via another interface).

#4
It's a bit unclear to me what you are trying to achieve. Matter is an application protocol. Matter devices are registered to a Matter controller. That's called a "fabric". The Matter controller usually happens to live on a border router. Your Dirigera hub is a border router that connects devices on Ethernet and Thread networks. So, if you want to communicate on the Matter level, connect to the controller via its name or IP, like you do with any other device on your local networks.

If however, you want to communicate to your Matter devices bypassing the controller, you will need to explicitly configure routes. But, as you already found out, you have no control over the ULAs. This, in fact, is by design. Thread is a radio network. If communication between devices on that network breaks down for whatever reason, the network will reorganise into so called partitions, i.e. partial networks where communication still works. The partitions will reconfigure and choose a new ULA. Same thing happens when partitions eventually rejoin.

To work around this simply add the Dirigera controller as a Matter hub to your Home Assistant, if there is support for this. That way, HA and Dirigera will talk Matter protocol. Don't communicate with your Matter devices via low-level IP, i.e. don't use multi-admin mode. It only drains batteries faster.

If that's not an option, follow Patrick's advice and put your HA into the same network as the Dirigera hub (possibly via another interface).
#5
Quote from: Monviech (Cedrik) on November 09, 2026, 06:09:23 PMI'm the maintainer of the ndp proxy btw, if you find something actionable I can help, but just a logging issue isn't actionable from my side right now.
I didn't mean to push you or anyone. I'm looking for pointers because I'm at a loss how to proceed investigating other than reading the code.

For example, there are a number of automatic rules that would fit the logged messages, but they are "ipv6-icmp from (self) to fe80::/10 icmp6-type neighbrsol", etc. Is there a chance that packets forwarded by the proxy match "(self)" even though the actual source address is not one of the firewall's?
#6
After some more investigation it appears that the firewall messages from unknown origin are indeed related to the NDP proxy. Turning the NDP proxy on an interface off makes the messages go away, turn them on and the messages appear. As far as I can tell they are actual NDP messages from devices on the network on that interface, a virtual interface.

Still, I can't find a rule. I have tried this
root@firewall:~ # pfctl -s rules | grep '^pass out' | grep 'vlan03'
pass out quick on vlan03 proto udp from (self) port = bootps to any port = bootpc keep state label "226b7aa626e7457a57f778268b65f653"
pass out quick on vlan03 inet6 proto udp from (self) port = dhcpv6-server to fe80::/10 keep state label "b5e319b6fe5505c953c8ffb4303e737c"
as the messages are "out" and on interface vlan03, but none matches IPV6-ICMP.

Without reading the code: Is there a hack in NDP proxy that inserts firewall messages directly into whatever the firewall live view reads from?
#7
It just realised that the routes entered by NDP proxy don't go away when an interface is removed from the config. To me, that's a clear indication that the communication is expected to stop and the routes should go away. But I agree that routes should not be removed when stopping / restarting the proxy.

Not terribly important though ...
#8
Today I noticed that the firewall log of one of my OPNsense 26.10-amd64 FWs contains messages as shown in the screenshot. The label field and rid are empty, just like messages from rules that had their logging turned off afterwards. But these continue to be produced at ~1h intervals and I have failed to identify a matching rule, manual or automatic, with or without logging.

I noticed however, that these messages only appear on 2 VLANS that also happen to use os-ndp-proxy-go. I had the chance to disable the proxy on Skynet and the messages stopped. They returned when the proxy was re-activated. I don't consider this a proof and will investigate this further.
#9
There's a minor problem in the rules UI: Without any interface rules (on MyLAN) the message "No MyLAN rules have been defined. All incoming connections on this interface will be blocked until you add a pass rule. Exceptions for automatically generated or floating rules may apply." is shown. To be perfectly correct, the message should also mention group rules. Actually, I think the message should not pop up at all. Block by default is fundamental to OPNsense.
#10
The CRL issue can be resolved by choosing "Deciso (HTTPS,NL,Commercial)" as the mirror instead of "Deciso (HTTPS,Commercial)"
#11
26.7, 26.10 Series / Fresh 26.10 installation report
October 30, 2026, 03:55:40 PM
Today, I did a fresh install on a DEC750. The machine had a bit of a checkered past. So, I wanted to take the opportunity to start from scratch, install the latest BIOS and do away with the legacy stuff, moving to dnsmasq and the new rules at the same time. Everything had been prepared on a test machine running 26.7.x and that backup was used to restore the configuration of the DEC750.

In summary, everything went well.

Some things didn't work as expected though:

  • I chose the OPNsense-business-26.10-vga-amd64 installation image. I never got a login prompt on the console during installation. I probably should have chosen the serial version. Login via ssh didn't work either because
  • The LAN interface does not issue dhcp addresses by default. I had to configure the ethernet interface of my PC manually.
  • The installation and restoring the config went just fine. However, a check for updates after the reboot resulted in the re-installation of the OPNBEcore package. Not sure if this is the intended behaviour after entering the subscription key. In any case, there were a number of messages about missing CRLs.

Here's the log output:

***GOT REQUEST TO UPDATE***
Currently running OPNsense 26.10 (amd64) at Fri Oct 30 15:27:31 CET 2026
Strict TLS 1.3 and CRL checking is enabled.
Updating OPNsense repository catalogue...
No CRL was provided for CA /C=AT/O=ZeroSSL GmbH/CN=ZeroSSL RSA DV SSL CA 2
No CRL was provided for CA /C=GB/O=Sectigo Limited/CN=Sectigo Public Server Authentication Root R46
No CRL was provided for CA /C=AT/O=ZeroSSL GmbH/CN=ZeroSSL RSA DV SSL CA 2
No CRL was provided for CA /C=GB/O=Sectigo Limited/CN=Sectigo Public Server Authentication Root R46
OPNsense repository is up to date.
All repositories are up to date.
Updating OPNsense repository catalogue...
No CRL was provided for CA /C=AT/O=ZeroSSL GmbH/CN=ZeroSSL RSA DV SSL CA 2
No CRL was provided for CA /C=GB/O=Sectigo Limited/CN=Sectigo Public Server Authentication Root R46
No CRL was provided for CA /C=AT/O=ZeroSSL GmbH/CN=ZeroSSL RSA DV SSL CA 2
No CRL was provided for CA /C=GB/O=Sectigo Limited/CN=Sectigo Public Server Authentication Root R46
OPNsense repository is up to date.
All repositories are up to date.
Checking for upgrades (166 candidates): .......... done
Processing candidates (166 candidates): . done
The following 1 package(s) will be affected (of 0 checked):

Installed packages to be REINSTALLED:
opnsense-business-26.10 [OPNsense] (Vital flag changed: 'true' -> 'false')

Number of packages to be reinstalled: 1

6 MiB to be downloaded.
No CRL was provided for CA /C=AT/O=ZeroSSL GmbH/CN=ZeroSSL RSA DV SSL CA 2
No CRL was provided for CA /C=GB/O=Sectigo Limited/CN=Sectigo Public Server Authentication Root R46
[1/1] Fetching opnsense-business-26.10: .......... done
Checking integrity... done (0 conflicting)
[1/1] Reinstalling opnsense-business-26.10...
[1/1] Extracting opnsense-business-26.10: .......... done
Stopping configd...done
Resetting root shell
Updating /etc/shells
Unhooking from /etc/rc
Unhooking from /etc/rc.shutdown
Updating /etc/shells
Registering root shell
Hooking into /etc/rc
Hooking into /etc/rc.shutdown
Starting configd.
>>> Invoking update script 'refresh.sh'
Flushing all caches...done.
Writing firmware settings: OPNsense
Writing trust files...done.
Scanning /usr/share/certs/untrusted for certificates...
Scanning /usr/share/certs/trusted for certificates...
Scanning /usr/local/share/certs for certificates...
certctl: No changes to trust store were made.
Writing trust bundles...done.
Configuring login behaviour...done.
Configuring cron...done.
Configuring system logging...done.
=====
Message from opnsense-business-26.10:

--
Just a dream and the wind to carry me
Checking integrity... done (0 conflicting)
Nothing to do.
Checking all packages: .......... done
The following package files will be deleted:
/var/cache/pkg/opnsense-business-26.10.pkg
/var/cache/pkg/opnsense-business-26.10~2$gokh8ff6.pkg
The cleanup will free 6 MiB
Deleting files: .. done
Nothing to do.
Flushing temporary package files... done
Starting web GUI...done.
***DONE***
#12
Thanks to all for the very valuable suggestions. I will look into the multi-interface group membership concept to compose rule sets more.
#13
Quote from: Bob.Dig on October 22, 2026, 06:26:22 PMI wonder, if I could also do some more rules on such a group like block RFC1918, allow internet and do those as non-quick. Then I could create some local quick allow rules on the single interfaces... 
Yes, that can be done. This how I implemented traffic rules for multiple user groups.
#14
I'm planning a network with >100 LAN networks on VLANs. Every network needs access to DNS, NTP and possibly other services on the firewall.

The easy solution is of course, to create a floating rule, so that any firewall interface except WAN(s) can be reached from any network. But that would allow nosy users and internal attackers to easily get an idea of the topology of the network. Alternatively, creating more specialised rules for every interface would be a pain. The best I have come up with so far is to create an otherwise unused interface, have all the firewall services listen on that and to create one floating rule for everyone to access the services there. That adds a bit of fiddling with DHCP.

Is there a more elegant way, i.e. secure and low maintenance, to provide services for every network on the FW yet keep the networks seperate?
#15
German - Deutsch / Re: [Hardware] Temperaturen
September 29, 2026, 03:28:13 PM
Ich würde mit dem smartctl plugin mal nachsehen wie die SSD gebraucht wird. Wichtig ist die Info bei "Data Units Written:". Daraus kann man dann grob die noch verbleibende Lebensdauer der SSD abschätzen. Im Zweifelsfall RRD abschalten und unter Systemeinstellungen die RAM Disk aktivieren, um Schreibzugriffe zu minimieren. Unter "Available Spare" sollte immer 100% stehen, andernfalls ist das Ende schon in Sicht.

Unter "Supported LBA Sizes" wird die "natürliche" Blockgröße der SDD durch ein "+" angezeigt. Bei vielen SSDs ist aus Kompatibilitätsgründen jedoch gerne mal 512 Bytes eingestellt, obwohl 4096 die bessere Wahl wäre. Insbesondere in Kombination mit ZFS als Dateisystem sollte man das mal durchdenken, damit Write-Amplification vermieden wird. Das ist ein relativ weites Feld, das ich hier nicht erklären möchte.

Smartctl zeigt außerdem in "Warning  Comp. Temperature Time:" und "Critical Comp. Temperature Time:" an, wie lange die SSD außerhalb ihrer Wohlfühlzone betrieben wurde.