Netgate Discussion Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search
    • Register
    • Login

    Crash when enabling traffic shaper on more than 1 port

    Scheduled Pinned Locked Moved 2.4 Development Snapshots
    27 Posts 8 Posters 2.8k Views
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • X
      xbipin
      last edited by

      @chrcoluk:

      I would test on 2.4.1.

      The igb driver on FreeBSD 11.0 (pfSense 2,4) is missing 'hw.igb.buf_ring_size' tunable so setting it on 2.4 probably has no affect, and igb had some bugs which are fixed in FreeBSD 11.1 (pfSense 2.4.1.). The sysctl 'hw.igb.buf_ring_size' is also reinstated.

      isnt it possible to backport the igb drivers from 11.1 for now rather than ship pfsense 2.4 with broken traffic shaping?

      1 Reply Last reply Reply Quote 0
      • dotOneD
        dotOne
        last edited by

        Did someone test the traffic shaping on 2.4.1?
        I didn’t have the time to do it this week, and next week I don’t have the time either.

        1 Reply Last reply Reply Quote 0
        • C
          chrcoluk
          last edited by

          @avink:

          Did someone test the traffic shaping on 2.4.1?
          I didn’t have the time to do it this week, and next week I don’t have the time either.

          it needs you guys with the problem to test, no point in me testing as I dont get the crash on 2.4 anyway.

          Its far easier for you to test than for someone to backport the driver, (note the fixes are also in the kernel rather than the driver).

          Do a config backup.
          Download 2.4.1 snapshot from pfsense servers.
          Install 2.4.1
          Restore the backup.

          pfSense CE 2.7.2

          1 Reply Last reply Reply Quote 0
          • X
            xbipin
            last edited by

            but the main question is how stable is 2.4.1 coz on my box on 2.4 it doesnt crash and on this i can test but where it crashes is a client box in a remote location so if 2.4.1 isnt stable then it will be cause a lot of headache for me bringing it back to 2.4

            1 Reply Last reply Reply Quote 0
            • P
              PiBa
              last edited by

              I tried 2.4.1 with 'em' E1000 nics on ESXi.. for me it crashes..

              Version	2.4.1-DEVELOPMENT (amd64) 
built on Sat Sep 23 19:46:59 CDT 2017 
FreeBSD 11.1-RELEASE-p1 


              amd64
11.1-RELEASE-p1
FreeBSD 11.1-RELEASE-p1 #299 r313908+13ee0afae40(RELENG_2_4): Sat Sep 23 19:54:08 CDT 2017     root@buildbot2.netgate.com:/builder/ce/tmp/obj/builder/ce/tmp/FreeBSD-src/sys/pfSense

Crash report details:

No PHP errors found.

Filename: /var/crash/bounds
1

Filename: /var/crash/info.0
Dump header from device: /dev/da0s1b
  Architecture: amd64
  Architecture Version: 1
  Dump Length: 86016
  Blocksize: 512
  Dumptime: Sun Sep 24 19:07:58 2017
  Hostname: pfsense_3.local
  Magic: FreeBSD Text Dump
  Version String: FreeBSD 11.1-RELEASE-p1 #299 r313908+13ee0afae40(RELENG_2_4): Sat Sep 23 19:54:08 CDT 2017
    root@buildbot2.netgate.com:/builder/ce/tmp/obj/builder/ce/tmp/FreeBSD-src/sys/pfSense
  Panic String: hfsc_dequeue:
  Dump Parity: 1670814533
  Bounds: 0
  Dump Status: good

Filename: /var/crash/info.last
Dump header from device: /dev/da0s1b
  Architecture: amd64
  Architecture Version: 1
  Dump Length: 86016
  Blocksize: 512
  Dumptime: Sun Sep 24 19:07:58 2017
  Hostname: pfsense_3.local
  Magic: FreeBSD Text Dump
  Version String: FreeBSD 11.1-RELEASE-p1 #299 r313908+13ee0afae40(RELENG_2_4): Sat Sep 23 19:54:08 CDT 2017
    root@buildbot2.netgate.com:/builder/ce/tmp/obj/builder/ce/tmp/FreeBSD-src/sys/pfSense
  Panic String: hfsc_dequeue:
  Dump Parity: 1670814533
  Bounds: 0
  Dump Status: good

Filename: /var/crash/minfree
2048

Filename: /var/crash/textdump.tar.0
ddb.txt06000014000013161763156  7103 ustarrootwheeldb:0:kdb.enter.default>  run lockinfo
db:1:lockinfo> show locks
No such command
db:1:locks>  show alllocks
No such command
db:1:alllocks>  show lockedvnods
Locked vnodes
db:0:kdb.enter.default>  show pcpu
cpuid        = 1
dynamic pcpu = 0xfffffe00981a3200
curthread    = 0xfffff80003215560: pid 0 "em4 taskq"
curpcb       = 0xfffffe002b4f4b80
fpcurthread  = none
idlethread   = 0xfffff800031df560: tid 100004 "idle: cpu1"
curpmap      = 0xffffffff82a3a7c0
tssp         = 0xffffffff82a738f8
commontssp   = 0xffffffff82a738f8
rsp0         = 0xfffffe002b4f4b80
gs32p        = 0xffffffff82a7a150
ldt          = 0xffffffff82a7a190
tss          = 0xffffffff82a7a180
db:0:kdb.enter.default>  bt
Tracing pid 0 tid 100036 td 0xfffff80003215560
kdb_enter() at kdb_enter+0x3b/frame 0xfffffe002b4f3cd0
vpanic() at vpanic+0x1a3/frame 0xfffffe002b4f3d50
panic() at panic+0x43/frame 0xfffffe002b4f3db0
hfsc_dequeue() at hfsc_dequeue+0x1a6/frame 0xfffffe002b4f3df0
tbr_dequeue() at tbr_dequeue+0xea/frame 0xfffffe002b4f3e40
vlan_start() at vlan_start+0x196/frame 0xfffffe002b4f3e90
if_transmit() at if_transmit+0x15c/frame 0xfffffe002b4f3ed0
ether_output() at ether_output+0x718/frame 0xfffffe002b4f3f60
ip_output() at ip_output+0x154a/frame 0xfffffe002b4f40b0
ip_forward() at ip_forward+0x323/frame 0xfffffe002b4f4150
ip_input() at ip_input+0x75a/frame 0xfffffe002b4f41b0
netisr_dispatch_src() at netisr_dispatch_src+0xa0/frame 0xfffffe002b4f4200
ng_iface_rcvdata() at ng_iface_rcvdata+0x11a/frame 0xfffffe002b4f4230
ng_apply_item() at ng_apply_item+0xde/frame 0xfffffe002b4f42b0
ng_snd_item() at ng_snd_item+0x18c/frame 0xfffffe002b4f42f0
ng_apply_item() at ng_apply_item+0xde/frame 0xfffffe002b4f4370
ng_snd_item() at ng_snd_item+0x18c/frame 0xfffffe002b4f43b0
ng_ppp_rcvdata() at ng_ppp_rcvdata+0xe6b/frame 0xfffffe002b4f4430
ng_apply_item() at ng_apply_item+0xde/frame 0xfffffe002b4f44b0
ng_snd_item() at ng_snd_item+0x18c/frame 0xfffffe002b4f44f0
ng_apply_item() at ng_apply_item+0xde/frame 0xfffffe002b4f4570
ng_snd_item() at ng_snd_item+0x18c/frame 0xfffffe002b4f45b0
ng_apply_item() at ng_apply_item+0xde/frame 0xfffffe002b4f4630
ng_snd_item() at ng_snd_item+0x18c/frame 0xfffffe002b4f4670
ether_demux() at ether_demux+0x240/frame 0xfffffe002b4f46a0
ether_nh_input() at ether_nh_input+0x310/frame 0xfffffe002b4f4700
netisr_dispatch_src() at netisr_dispatch_src+0xa0/frame 0xfffffe002b4f4750
ether_input() at ether_input+0x26/frame 0xfffffe002b4f4770
vlan_input() at vlan_input+0x1f0/frame 0xfffffe002b4f47f0
ether_demux() at ether_demux+0x156/frame 0xfffffe002b4f4820
ether_nh_input() at ether_nh_input+0x310/frame 0xfffffe002b4f4880
netisr_dispatch_src() at netisr_dispatch_src+0xa0/frame 0xfffffe002b4f48d0
ether_input() at ether_input+0x26/frame 0xfffffe002b4f48f0
if_input() at if_input+0xa/frame 0xfffffe002b4f4900
lem_rxeof() at lem_rxeof+0x3df/frame 0xfffffe002b4f49a0
lem_handle_rxtx() at lem_handle_rxtx+0x32/frame 0xfffffe002b4f49e0
taskqueue_run_locked() at taskqueue_run_locked+0x127/frame 0xfffffe002b4f4a40
taskqueue_thread_loop() at taskqueue_thread_loop+0xc8/frame 0xfffffe002b4f4a70
fork_exit() at fork_exit+0x85/frame 0xfffffe002b4f4ab0
fork_trampoline() at fork_trampoline+0xe/frame 0xfffffe002b4f4ab0
--- trap 0, rip = 0, rsp = 0, rbp = 0 ---
              1 Reply Last reply Reply Quote 0
              • D
                doktornotor Banned
                last edited by

                Afraid this is really an uphill battle and altq is pretty much dead… Not sure what's the plan in FreeBSD but this buggy mess is not maintainable by Netgate alone.

                1 Reply Last reply Reply Quote 0
                • C
                  chrcoluk
                  last edited by

                  Yes its reliant on FreeBSD upstream, pfSense may refine the code, but you entering a dangerous situation if you start making large en masse changes from the upstream product as this would need maintaining outside of upstream and be a pain on every upstream update.

                  FreeBSD made a decision to not keep up with openbsd in regards to PF and ALTQ so they made the rod on their own back with that decision, the good news is that they have started to fix FreeBSD specific PF issues now and I would hope this includes ALTQ as well.

                  The problem in this case there is no specific confirmed bug, just a few guys reporting crashes at high load, it needs lots of testing to dig deeper.

                  My advice was on the basis I am aware of scheduler panic fixes and igb panic fixes in FreeBSD 11.1 upstream hence what I posted.

                  pfSense CE 2.7.2

                  1 Reply Last reply Reply Quote 0
                  • D
                    doktornotor Banned
                    last edited by

                    @chrcoluk:

                    FreeBSD made a decision to not keep up with openbsd in regards to PF and ALTQ so they made the rod on their own back with that decision, the good news is that they have started to fix FreeBSD specific PF issues now and I would hope this includes ALTQ as well.

                    ALTQ has been gone from OpenBSD for years. http://undeadly.org/cgi?action=article&sid=20140419151959

                    1 Reply Last reply Reply Quote 0
                    • X
                      xbipin
                      last edited by

                      my crash log attached as follows, im also having same issue on client box when i enable shaping, sorry the log was too long to capture completely

                      crash.txt
                      crash2.txt

                      1 Reply Last reply Reply Quote 0
                      • dotOneD
                        dotOne
                        last edited by

                        A quick update from my side.
                        Both ALTQ and HFSC crash on 2.4.1 as well.
                        It doesn’t have to have any load. It crashes only seconds after reboot.

                        Will try to upload the crash info tonight.

                        1 Reply Last reply Reply Quote 0
                        • C
                          chrcoluk
                          last edited by

                          @doktornotor:

                          @chrcoluk:

                          FreeBSD made a decision to not keep up with openbsd in regards to PF and ALTQ so they made the rod on their own back with that decision, the good news is that they have started to fix FreeBSD specific PF issues now and I would hope this includes ALTQ as well.

                          ALTQ has been gone from OpenBSD for years. http://undeadly.org/cgi?action=article&sid=20140419151959

                          It is gone now yes, but before it was removed, it was updated and improved over what the version in FreeBSD is.

                          pfSense CE 2.7.2

                          1 Reply Last reply Reply Quote 0
                          • P
                            PiBa
                            last edited by

                            Luiz fixed it.!. ;D

                            https://redmine.pfsense.org/issues/7879#note-3

                            At least after initial testing with and pushing some traffic through my queue's which seemed like a sure way to crash it before, i'm gone try and see how it holds up tomorrow :)

                            2.4.0-RC (amd64) 
built on Mon Sep 25 15:04:23 CDT 2017 
FreeBSD 11.0-RELEASE-p12

                            Perhaps some of you guy's here can also give it another testrun.? 8)

                            1 Reply Last reply Reply Quote 0
                            • X
                              xbipin
                              last edited by

                              i tried on 2 apu2 boxes and now works all good, can any1 point me to the actual commit that fixed this?

                              1 Reply Last reply Reply Quote 0
                              • dotOneD
                                dotOne
                                last edited by

                                It's running for a few hours now.
                                No crash since the reboot, so this looks promising.

                                1 Reply Last reply Reply Quote 0
                                • First post
                                  Last post
                                Copyright 2025 Rubicon Communications LLC (Netgate). All rights reserved.