Now Available: pfSense® Plus 25.07-RELEASE

[SteveITS]

@crucialguy That’s a bug in 24.* but if you go to the config history page it will prune them out, but the page will probably time out before it does.

[markster]

After upgrade I noticed few things that stopped working for me.

1. Avahi logs show that all has started fine but apple mirroring or airplay does not work anymore. System logs shows that Ahavi started on selected subnets but I do not think that it makes necessary changes to IP tables and enable cross subnet communication.

2. Enabling KEA DHCP works but some LG TV clients would not "auto connect". (my Unbound DNS is on different host). To get it to connect I have to disable "auto connect" and configure TV to use static IP.

[dennypage]

@markster said in Now Available: pfSense Plus 25.07-RELEASE:

System logs shows that Ahavi started on selected subnets but I do not think that it makes necessary changes to IP tables and enable cross subnet communication.

Avahi does not make firewall rule changes, and no changes are necessary for Avahi to function with the default set of firewall rules. This has been the case for many, many years.

It's possible to break Avahi functionality If you specifically add rules to block inbound mDNS traffic to firewall interfaces, but that would be a conscious act.

[johnpoz]

@dennypage since I know airprint sends to multicast address 224.0.0.251, I would assume airplay does the same for discovery via mdns - I have never bothered to look that deep into them when did testing bfore for other threads on how to validate mdns is being passed on. But options might be set. Which would be blocked now unless rule set to allow them.

If that is the case - I would have to test, but its late here - then yeah it could be blocked now without conscious act to allow.

But yeah normally if the interface has an allow rule for internet which normally destination any avahi would pass on the mdns to where you set it to go. But with multicast with options set being blocked in recent versions, maybe you would now have to allow for options in your allow rule?

I can test in the morning - I am not a fan or user of avahi, not really a fan of breaking L2 barrier.. I just put my printer that uses airprint on the network my wifi clients use that would want to airprint to that printer. Still not a fan of apple not letting you just setting an IP on ios vs having to discover the printer. My other devices that need to talk to printer from other networks, just set the IP/FQDN of the printer to print to it.

But if it use to work, and now doesn't after an upgrade to pfsense the options on multicast would be my guess.

[dennypage]

@johnpoz said in Now Available: pfSense Plus 25.07-RELEASE:

@dennypage since I know airprint sends to multicast address 224.0.0.251, I would assume airplay does the same for discovery via mdns - I have never bothered to look that deep into them when did testing bfore for other threads on how to validate mdns is being passed on. But options might be set. Which would be blocked now unless rule set to allow them.
If that is the case - I would have to test, but its late here - then yeah it could be blocked now without conscious act to allow.

mDNS packets do not have options set. And remember, by default pfSense has always dropped packets with options set — it’s just that the drops were silent until 24.x.

The issue regarding multicast with options is with IGMP, which requires the router alert option.

[johnpoz]

@dennypage true - they did always drop before. When I have tested this I always created a floating to rule to make sure 224.0.0.251 was for sure allowed. But all good points.

[markster]

@dennypage I did not made any changes just enable Avahi on the 2 interfaces. It is always been like that.
Anyway, I decided to make things easier without Avahi and place both TV's on the same segment as my AP router. AP router has 4 Ethernet ports so it can act as AP and a switch. All working good.

[johnpoz]

@markster said in Now Available: pfSense Plus 25.07-RELEASE:

to make things easier without Avahi and place both TV's on the same segment as my AP router.

There you go.. I personally think that is a much better solution - and requires no breaking of the L2 barrier. mdns was never meant to cross networking boundaries.

[jos-andel]

Upgraded from 24.11 to 25.07 on our spare Netgate 7100 unit. We always do this for testing, and so far it looks good.

But after the upgrade monitoring complains:
SWAP CRITICAL - 0% free (0 MB out of 0 MB) - Swap is either disabled, not present, or of zero size.

And monitoring is probably right, since there seems to be no more swapspace after the upgrade:
[25.07-RELEASE][admin@spare7100]/root: swapinfo
Device 1K-blocks Used Avail Capacity

On my other units which still are on 24.11 I get:
/dev/ada0p2 1048576 0 1048576 0%
/dev/ada1p2 1048576 0 1048576 0%

On my upgraded box /etc/fstab still has the entries for swapspace:

Device Mountpoint FStype Options Dump Pass#

/dev/ada0p2 none swap sw 0 0
/dev/ada1p2 none swap sw 0 0

When I run 'swapon -a' on command-line, the swap activates and monitoring turns green once more.

Is this intended to be this way, or a bug maybe, or just bad luck this boot ?

Tomorrow I can try another reboot, see what happens then.

Jos Andel

[stephenw10]

Nope that's not expected. SWAP should be enabled by default if it's present. Are you running UFS or ZFS?

[jos-andel]

@stephenw10
We are running on zfs. Today I rebooted to test if I could reproduce it, after this reboot still no swapspace and another alert in the monitoring. So looks like reproducable.

Thanks, Jos

[stephenw10]

Do you see any errors at boot? Have you ever made any custom changes to the SWAP setup?

[jos-andel]

@stephenw10
We never changed swap setup. Always upgraded through the normal upgrade procedure. From version 23 to 25.

In my dmesg boot output:
module_register_init: MOD_LOAD (ipw_bss_fw, 0xffffffff80760600, 0) error 1
module_register_init: MOD_LOAD (ipw_ibss_fw, 0xffffffff807606b0, 0) error 1
module_register_init: MOD_LOAD (ipw_monitor_fw, 0xffffffff80760760, 0) error 1
module_register_init: MOD_LOAD (iwi_bss_fw, 0xffffffff8077fdd0, 0) error 1
module_register_init: MOD_LOAD (iwi_ibss_fw, 0xffffffff8077fe80, 0) error 1
module_register_init: MOD_LOAD (iwi_monitor_fw, 0xffffffff8077ff30, 0) error 1

But nothing related to swap.

[stephenw10]

Yeah those errors from the wifi drivers are normal when the hardware isn't present.

Hmm, I wonder if this is because you have multiple SWAP devices.

Try running: /bin/cat /etc/fstab | /usr/bin/grep swap | /usr/bin/awk '{ printf $1 }' | /usr/bin/head -n 1

That's what the RC script runs to get the swap device, if it returns something unexpected it may fail.

Do you see /tmp/fstab.swap present?

[jos-andel]

@stephenw10
I think you're onto something.

25.07-RELEASE][admin@vpnspare.mgmt.blah]/root: /bin/cat /etc/fstab | /usr/bin/grep swap | /usr/bin/awk '{ printf $1 }' | /usr/bin/head -n 1
/dev/ada0p2/dev/ada1p2[25.07-RELEASE][admin@vpnspare.mgmt.blah]/root:

cat /tmp/fstab.swap
/dev/ada0p2/dev/ada1p2 none swap sw 0 0

That doesn't look like a correct device, so it's quite likely that this breaks it. We have 4 Netgate 7100's which are like this, with two SSD's built in. We got them delivered like this, no modifications done (except for a broken PSU two times).

On 24.11 it works just fine, and probably on all previous versions we used as well in the past 3 years or so.

On 24.11 it looks like this:
/root: swapinfo
Device 1K-blocks Used Avail Capacity
/dev/ada0p2 1048576 0 1048576 0%
/dev/ada1p2 1048576 0 1048576 0%
Total 2097152 0 2097152 0%

root: cat /tmp/fstab.swap

Device Mountpoint FStype Options Dump Pass#

/dev/ada0p2 none swap sw 0 0
/dev/ada1p2 none swap sw 0 0

Thanks, Jos

[stephenw10]

Aha! Ok let me do some digging here....

[stephenw10]

Ah, already reported in 2.8: https://redmine.pfsense.org/issues/16232

Looks like an easy enough fix though. Patch incoming....

[stephenw10]

Ok this works for me: 175.diff

It's a very simple (one character!) change so you might just manually edit the file. Let us know if you're able to test that.

[jos-andel]

@stephenw10

Just tested the patched pfSense-rc on our spare unit.

/etc: diff pfSense-rc.org pfSense-rc
590c590
< SWAPDEVICE=/bin/cat /etc/fstab | /usr/bin/grep swap | /usr/bin/awk '{ printf $1 }' | /usr/bin/head -n 1

SWAPDEVICE=/bin/cat /etc/fstab | /usr/bin/grep swap | /usr/bin/awk '{ print $1 }' | /usr/bin/head -n 1

Works like a charm:
: swapinfo
Device 1K-blocks Used Avail Capacity
/dev/ada0p2 1048576 0 1048576 0%
/dev/ada1p2 1048576 0 1048576 0%
Total 2097152 0 2097152 0%

: cat /tmp/fstab.swap

Device Mountpoint FStype Options Dump Pass#

/dev/ada0p2 none swap sw 0 0
/dev/ada1p2 none swap sw 0 0
[25.07-RELEASE][admin@arn1-ops-vpnspare.mgmt.sidn.nl]/etc:

[mvikman]

@stephenw10

I'm planning to upgrade, but still on 24.11 and I have ZFS mirror with GEOM mirrored swap, would this problem affect my system or is it just for systems with multiple separate (non-mirrored) swap partitions on the both drives?