Build pfSense-Full-Update-2.0-BETA4-20101109-0016.tgz seems broken
-
I upgraded to pfSense-Full-Update-2.0-BETA4-20101109-0016.tgz from a build of early October (probably pfSense-Full-Update-2.0-BETA4-20101002-1627.tgz). Before upgrading I made the IPv6 changes (gitsync to upgrade the GUI) and configured a tunnel for IPv6.
With LAN and WAN both plugged in and running after completion of the upgrade I had a crash with apinger the current process. I removed the cables from the WAN and LAN interfaces and had a panic syncache: mbuf too small
I don't know if the IPv6 related changes are a factor in the crash and panic.
-
I tar'd up my production pfSense system (version 1.2.3) and restored to the "V2", upgraded firmware to pfSense-Full-Update-2.0-BETA4-20101110-0504.tgz and restored the previous v2 config and haven't seen the panic again. I'll try the IPv6 gitsync within the next 12 hours and report on that.
There seems to be a DNS forwarder problem with pfSense-Full-Update-2.0-BETA4-20101110-0504.tgz which I'm still looking into.
-
There seems to be a DNS forwarder problem with pfSense-Full-Update-2.0-BETA4-20101110-0504.tgz which I'm still looking into.
rl0 interface to the DNS was "stuck" after reporting watchdog timeout many times. Removing the cable and putting it back in seemed to break it out of that state and DNS now works.
-
I have seen the mbuf too small panic but only on a dev kernel.
Were you running a normal kernel?
The mbuf size was increased on amd64, it may need increased on i386 as well. Ermal was waiting for more input to find out what the problem might be.
-
I think I recall seeing some recent commits with changes to mbufs (not just that mbuf size parameter) and to something that uses them. Could that have anything to do with it?
-
Yes, hence the reason that adjusting the size fixed amd64. The patches were added a week or so ago to increase performance.
-
I've seen a number of page fault panics while attempting playback gitsync. The display disappears too quickly for me to capture. When attempting to get a kernel that would pause on panic I trashed something and had to repair my system. I used a cloned 1.2.3 system, upgraded the firmware to pfSense-Full-Update-2.0-BETA4-20101109-0016.tgz and restored my previous config. I still saw the page fault panic part way through playback gitsync, during the fetch phase.
I've been running normal kernels but will switch to a developers kernel to attempt to capture some more information on the page fault.
-
I switched to the developers kernel (# tar xzpf /kernels/kernel_Dev.gz -C /boot/ ) and rebooted but it didn't even complete the startup to console menu:
Fatal trap 12: page fault while in kernel mode
cpuid = 0; apic id = 00
fault virtual address = 0xdedeadc8
fault code = supervisor read, page not present
instruction pointer = 0x20:0xc0a51ce6
stack pointer = 0x28:0xc2763a88
frame pointer = 0x28:0xc2763a90
code segment = base 0x0, limit 0xfffff, type 0x1b
= DPL 0, pres 1, def32 1, gran 1
processor eflags = interrupt enabled, resume, IOPL = 0
current process = 12 (irq10: rl ehci0)