Kernel Panic

[jimp]

No - the hangs you can't get out of except by a power cycle. There is no crash dump from a hang, only from a panic. (This thread is really for the panics… the hangs have a separate thread)

[clahti]

ya that's what I thought, thanks.

[Slaygon]

Today one of the fw's (running "2.0-BETA5 (amd64) built on Thu Jan 27 19:46:43 EST 2011") managed to output something in /var/crash/ before it died.

See attached files.

ddb.txt
msgbuf.txt

[eri--]

That seems like another panic related to carp and your link going up down.
So seems like progress, at least to those who do not have intensive carp clusters but just hangs.
I will get back when i hav emore info on this new issue.

[Slaygon]

@ermal:

That seems like another panic related to carp and your link going up down.
So seems like progress, at least to those who do not have intensive carp clusters but just hangs.
I will get back when i hav emore info on this new issue.

I guess that would explain why the backup spontaneously becomes the master every now and then?
I get this on the backup:
BACKUP -> MASTER (preempting a slower master)

and this on the master:
MASTER -> BACKUP (more frequent advertisement received)

Adjusting advskew probably won't help here, I guess?

[eri--]

I have pushed some patches that should solve even the carp panic Slaygon.

So newest snapshots should have the fix for you.
I would skip the next one coming and get the other one.

[vito]

just tried snap
2.0-BETA5 (i386)
built on Fri Jan 28 05:30:15 EST 2011

and it crashed. :)

The only one i had any luck with is the first kernel that was posted.

Just so i am clear… the snaps have the correct kernel or do i still need to download from the separate link?

[Slaygon]

@ermal:

I have pushed some patches that should solve even the carp panic Slaygon.

So newest snapshots should have the fix for you.
I would skip the next one coming and get the other one.

So the "Fri Jan 28 05:30:15 EST 2011" is not the one you recommend for the carp fixes, but instead wait for the next one?
I'm currently on "Fri Jan 28 00:53:50 EST 2011".

Oh, and my backup just became the master spontaneously again.

Excellent work btw. Much appreciated!

[jimp]

yes, wait for the next one. It was just restarted to pick up the patches he pushed.

[eri--]

@vito please type a bt at that prompt next time.

[LostInIgnorance]

Loaded 2.0-BETA5 (i386) built on Fri Jan 28 05:30:15 EST 2011 running on a dell Intel(R) Pentium(R) 4 CPU 2.40GHz onboard gig nic (em)

Enter an option: Kernel page fault with the following non-sleepable locks held:
exclusive sleep mutex em0 (EM TX Lock) r = 0 (0xc2f52580) locked @ /usr/pfSensesrc/src/sys/dev/e1000/if_lem.c:1350
KDB: stack backtrace:
db_trace_self_wrapper(c0eb7ccb,ccc3ca90,c0a421c5,546,0,...) at db_trace_self_wrapper+0x26
kdb_backtrace(546,0,ffffffff,c145df04,ccc3cac8,...) at kdb_backtrace+0x29
_witness_debugger(c0eba1e3,ccc3cadc,4,1,0,...) at _witness_debugger+0x25
witness_warn(5,0,c0ef8592,2d05a8c0,c131be40,...) at witness_warn+0x20d
trap(ccc3cb68) at trap+0x19e
calltrap() at calltrap+0x6
--- trap 0xc, eip = 0xc0a61478, esp = 0xccc3cba8, ebp = 0xccc3cbb8 ---
m_tag_delete(c2fef600,dedeadc0,c2fef600,c2fef600,ccc3cbf0,...) at m_tag_delete+0x48
m_tag_delete_chain(c2fef600,0,c0e6f12d,0,c2ed9720,...) at m_tag_delete_chain+0x3f
mb_dtor_mbuf(c2fef600,100,0,c0a42c18,df,...) at mb_dtor_mbuf+0x35
uma_zfree_arg(c1d7e380,c2fef600,0,72,ccc3cc84,...) at uma_zfree_arg+0x29
m_freem(c2fef600,4,c0e6f12d,b87,c2f4e000,...) at m_freem+0x43
lem_txeof(c2f52580,0,c0e6f12d,546,c2f525bc,...) at lem_txeof+0x158
lem_handle_rxtx(c2f4e000,1,c0eb959c,4f,c2edb8d8,...) at lem_handle_rxtx+0x60
taskqueue_run(c2edb8c0,c2edb8d8,c0ea6955,0,c0eb2bfb,...) at taskqueue_run+0x103
taskqueue_thread_loop(c2f525ec,ccc3cd38,c0eaf76a,344,c131be40,...) at taskqueue_thread_loop+0x68
fork_exit(c0a3b440,c2f525ec,ccc3cd38) at fork_exit+0xb8
fork_trampoline() at fork_trampoline+0x8
--- trap 0, eip = 0, esp = 0xccc3cd70, ebp = 0 ---

Fatal trap 12: page fault while in kernel mode
cpuid = 0; apic id = 00
fault virtual address   = 0xdedeadc0
fault code              = supervisor read, page not present
instruction pointer     = 0x20:0xc0a61478
stack pointer           = 0x28:0xccc3cba8
frame pointer           = 0x28:0xccc3cbb8
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         = 0 (em0 taskq)
[thread]
Stopped at      m_tag_delete+0x48:      movl    0(%ecx),%eax
db> bt
Tracing pid 0 tid 64050 td 0xc2f4bc80
m_tag_delete(c2fef600,dedeadc0,c2fef600,c2fef600,ccc3cbf0,...) at m_tag_delete+0x48
m_tag_delete_chain(c2fef600,0,c0e6f12d,0,c2ed9720,...) at m_tag_delete_chain+0x3f
mb_dtor_mbuf(c2fef600,100,0,c0a42c18,df,...) at mb_dtor_mbuf+0x35
uma_zfree_arg(c1d7e380,c2fef600,0,72,ccc3cc84,...) at uma_zfree_arg+0x29
m_freem(c2fef600,4,c0e6f12d,b87,c2f4e000,...) at m_freem+0x43
lem_txeof(c2f52580,0,c0e6f12d,546,c2f525bc,...) at lem_txeof+0x158
lem_handle_rxtx(c2f4e000,1,c0eb959c,4f,c2edb8d8,...) at lem_handle_rxtx+0x60
taskqueue_run(c2edb8c0,c2edb8d8,c0ea6955,0,c0eb2bfb,...) at taskqueue_run+0x103
taskqueue_thread_loop(c2f525ec,ccc3cd38,c0eaf76a,344,c131be40,...) at taskqueue_thread_loop+0x68
fork_exit(c0a3b440,c2f525ec,ccc3cd38) at fork_exit+0xb8
fork_trampoline() at fork_trampoline+0x8
--- trap 0, eip = 0, esp = 0xccc3cd70, ebp = 0 ---
db>

Happened instantly when I tried to OpenVPN in[/thread]

[vito]

it is a hard lock…can't type anything in.
On reboot, still do not see the crash files that jimp added.

[jimp]

@vito:

it is a hard lock…can't type anything in.
On reboot, still do not see the crash files that jimp added.

As I said before, those won't be created for hard locks, only for panics/crashes.

[Slaygon]

Just seen this come out: "Fri Jan 28 13:06:21 EST 2011"…
Is this the one where the carp sync problems are addressed?

If so, it will take me a number of hours to get this applied and verified. 1-4 hours of active monitoring before the bug usually appears, however, I will not be able to start testing all too soon. Expecting about 2 to 8 hours before I can start monitoring.

If anyone could start testing sooner, that would be great!

(for those that don't know, I run two DL-type HP servers with quad Intel gbit NICs in them)

And again, great work pf team! Really appreciated!

[geewhz01]

If the carp patches were also pushed to the AMD version 2.0-BETA5 (amd64)
built on Fri Jan 28 13:06:21 EST 2011 then it's not fixed for me.  As soon as I make any change and it syncs the backup firewall locks or panics.  It's not a hard lock though, it does bring me to a prompt but not seeing any of crash files in the /var/crash.

Andy

[Slaygon]

@geewhz01:

If the carp patches were also pushed to the AMD version 2.0-BETA5 (amd64)
built on Fri Jan 28 13:06:21 EST 2011 then it's not fixed for me.  As soon as I make any change and it syncs the backup firewall locks or panics.  It's not a hard lock though, it does bring me to a prompt but not seeing any of crash files in the /var/crash.

Andy

It's the amd64 version I am running too.
Not too good news there, Andy. Sorry to hear.
And usually, with the carp errors, there seems to be a freeze, rather than a crash, which will render no trace dumps.
Sometimes, very unusual though, I get a debug prompt at which I can get a backtrace. Mostly the box just hangs.

I guess the devs would appreciate any input they can get hold of, so if you do get to a prompt and can type something in, give "bt" (as in backtrace) a go and see if you could provide any more info.

Cheerio.

[jimp]

I just tried on my amd64 vm and when I force a manual panic (sysctl debug.kdb.panic=1) I get a textdump, and no db> prompt. I'm not sure why someone on a current snapshot update would still be getting left at a db> prompt when it crashes, they should be gathering the info and rebooting on their own.

Doesn't help with the hangs, though, but the hangs are a different problem (and a different thread :-)

[geewhz01]

Here is what I get when carp tries to sync.  This is running snapshot version amd64-20110128-0938.

Thanks,

Andy

[jimp]

That snap was before the last fixes went in. Try the newest one.

[geewhz01]

@jimp:

That snap was before the last fixes went in. Try the newest one.

pfSense-Full-Update-2.0-BETA5-amd64-20110128-0938.tgz 28-Jan-2011 13:09 99M

That's the latest I see on the snapshots server.  Am I looking in the wrong spot?  http://snapshots.pfsense.org/FreeBSD_RELENG_8_1/amd64/pfSense_HEAD/updates/?C=M;O=D