Yuriy Tsibizov
2006-Apr-19 12:15 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
I run a 6.1-RC (as of Apr, 14th) and after a day nothing can be sent from em
interface.
It is a filtering+accounting bridge-on-a-stick, with em interface connected to
3Com Switch 4200 100Mbit (not Gbit) port with 4 tagged vlans on it:
# This file now contains just the overrides from /etc/defaults/rc.conf.
# Please make all changes to this file, not to /etc/defaults/rc.conf.
hostname="XXXX.gfk.ru"
defaultrouter="10.X.X.X"
ifconfig_sis0="10.X.X.X/24"
ifconfig_em0="up"
cloned_interfaces="vlan0 vlan1 vlan2 vlan3 bridge0 bridge1"
ifconfig_vlan0="84.X.X.X/28 vlan 21 vlandev em0 up"
ifconfig_vlan1="vlan 22 vlandev em0 up"
ifconfig_bridge0="addm vlan0 addm vlan1 up"
ifconfig_vlan2="217.X.X.X/29 vlan 31 vlandev em0 up"
ifconfig_vlan3="vlan 32 vlandev em0 up"
ifconfig_bridge1="addm vlan2 addm vlan3 up"
syslogd_flags="-ss"
sshd_enable="YES"
ipsec_enable="YES"
pf_enable="YES"
pf_rules="/etc/pf.freeroute"
pflog_enable="YES"
and kernel is GENERIC with unused hardware removed and
####
options DEVICE_POLLING
options IPSEC
options IPSEC_ESP
options IPSEC_FILTERGIF
device vlan
device pf
device pflog
device pfsync
device if_bridge
# altq(9). Enable the base part of the hooks with the ALTQ option.
# also have the ALTQ_NOPCC option.
options ALTQ
options ALTQ_CBQ # Class Bases Queueing
options ALTQ_RED # Random Early Detection
options ALTQ_RIO # RED In/Out
options ALTQ_HFSC # Hierarchical Packet Scheduler
options ALTQ_CDNR # Traffic conditioner
options ALTQ_PRIQ # Priority Queueing
options ALTQ_NOPCC # Required for SMP build
#options ALTQ_DEBUG
added.
em0 is
em0@pci0:10:0: class=0x020000 card=0x002e8086 chip=0x100e8086 rev=0x02 hdr=0x00
vendor = 'Intel Corporation'
device = '82540EM Gigabit Ethernet Controller'
class = network
subclass = ethernet
em0: <Intel(R) PRO/1000 Network Connection Version - 3.2.18> port
0xb400-0xb43f mem 0xd7ea0000-0xd7ebffff,0xd7e80000-0xd7e9ffff irq 18 at device
10.0 on pci0
em0: [GIANT-LOCKED]
em0: Ethernet address: 00:07:e9:09:c2:3d
Polling was not used (I'll try to work with polling enabled on em0).
ALTQ is not configured, pf has all block rules logged and I can't see
anything relevant logged to pflog.
This machine also use netgraph netflow for traffic accounting on vlan
interfaces, but shutting down all netgraph nodes does not help.
Turning vlans, bridge and em down and up does not help too.
Traffic is received (I can see pairs of OSPF HELO messages from ISP router
logged on pflog0 exactly every 10 seconds), but it looks like nothing can be
sent.
Yuriy Tsibizov,
GfK RUS
Alexey Karagodov
2006-Apr-20 08:44 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
try enabling pormisc mode on vlans 19.04.06, Yuriy Tsibizov <Yuriy.Tsibizov@gfk.ru> ???????(?):> > I run a 6.1-RC (as of Apr, 14th) and after a day nothing can be sent from > em interface. > > It is a filtering+accounting bridge-on-a-stick, with em interface > connected to 3Com Switch 4200 100Mbit (not Gbit) port with 4 tagged vlans on > it: > > # This file now contains just the overrides from /etc/defaults/rc.conf. > # Please make all changes to this file, not to /etc/defaults/rc.conf. > hostname="XXXX.gfk.ru" > defaultrouter="10.X.X.X" > ifconfig_sis0="10.X.X.X/24" > ifconfig_em0="up" > cloned_interfaces="vlan0 vlan1 vlan2 vlan3 bridge0 bridge1" > ifconfig_vlan0="84.X.X.X/28 vlan 21 vlandev em0 up" > ifconfig_vlan1="vlan 22 vlandev em0 up" > ifconfig_bridge0="addm vlan0 addm vlan1 up" > ifconfig_vlan2="217.X.X.X/29 vlan 31 vlandev em0 up" > ifconfig_vlan3="vlan 32 vlandev em0 up" > ifconfig_bridge1="addm vlan2 addm vlan3 up" > syslogd_flags="-ss" > > sshd_enable="YES" > ipsec_enable="YES" > pf_enable="YES" > pf_rules="/etc/pf.freeroute" > pflog_enable="YES" > > and kernel is GENERIC with unused hardware removed and > > #### > options DEVICE_POLLING > options IPSEC > options IPSEC_ESP > options IPSEC_FILTERGIF > > device vlan > device pf > device pflog > device pfsync > device if_bridge > > # altq(9). Enable the base part of the hooks with the ALTQ option. > # also have the ALTQ_NOPCC option. > options ALTQ > options ALTQ_CBQ # Class Bases Queueing > options ALTQ_RED # Random Early Detection > options ALTQ_RIO # RED In/Out > options ALTQ_HFSC # Hierarchical Packet Scheduler > options ALTQ_CDNR # Traffic conditioner > options ALTQ_PRIQ # Priority Queueing > options ALTQ_NOPCC # Required for SMP build > #options ALTQ_DEBUG > > added. > > em0 is > em0@pci0:10:0: class=0x020000 card=0x002e8086 chip=0x100e8086 rev=0x02 > hdr=0x00 > vendor = 'Intel Corporation' > device = '82540EM Gigabit Ethernet Controller' > class = network > subclass = ethernet > > em0: <Intel(R) PRO/1000 Network Connection Version - 3.2.18> port > 0xb400-0xb43f mem 0xd7ea0000-0xd7ebffff,0xd7e80000-0xd7e9ffff irq 18 at > device 10.0 on pci0 > em0: [GIANT-LOCKED] > em0: Ethernet address: 00:07:e9:09:c2:3d > > Polling was not used (I'll try to work with polling enabled on em0). > > ALTQ is not configured, pf has all block rules logged and I can't see > anything relevant logged to pflog. > > This machine also use netgraph netflow for traffic accounting on vlan > interfaces, but shutting down all netgraph nodes does not help. > > Turning vlans, bridge and em down and up does not help too. > > Traffic is received (I can see pairs of OSPF HELO messages from ISP router > logged on pflog0 exactly every 10 seconds), but it looks like nothing can be > sent. > > Yuriy Tsibizov, > GfK RUS > > _______________________________________________ > freebsd-stable@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-stable > To unsubscribe, send any mail to "freebsd-stable-unsubscribe@freebsd.org" >
Yuriy Tsibizov
2006-Apr-20 09:07 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
>> 19.04.06, Yuriy Tsibizov <Yuriy.Tsibizov@gfk.ru> ???????(?): >> I run a 6.1-RC (as of Apr, 14th) and after a day nothing can be sent from em interface.>> It is a filtering+accounting bridge-on-a-stick, with em interface connected to 3Com >> Switch 4200 100Mbit (not Gbit) port with 4 tagged vlans on it:>> Polling was not used (I'll try to work with polling enabled on em0).Polling doesn't help too>> This machine also use netgraph netflow for traffic accounting on vlan interfaces, >> but shutting down all netgraph nodes does not help.>> Turning vlans, bridge and em down and up does not help too.>> Traffic is received (I can see pairs of OSPF HELO messages from ISP router logged >> on pflog0 exactly every 10 seconds), but it looks like nothing can be sent.> try enabling pormisc mode on vlansIt was enabled (it is enabled by default when you add interface to bridge, I think) It also does not explain why it work for almost a day and then stops. On switch I can see zero RX counters on this port (and non-zero TX). Yuriy.
Andrew Thompson
2006-Apr-20 10:55 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
On Wed, Apr 19, 2006 at 04:15:10PM +0400, Yuriy Tsibizov wrote:> I run a 6.1-RC (as of Apr, 14th) and after a day nothing can be sent from em interface. > > It is a filtering+accounting bridge-on-a-stick, with em interface > connected to 3Com Switch 4200 100Mbit (not Gbit) port with 4 tagged > vlans on it: > > Polling was not used (I'll try to work with polling enabled on em0). > > ALTQ is not configured, pf has all block rules logged and I can't see anything relevant logged to pflog. > > Traffic is received (I can see pairs of OSPF HELO messages from ISP > router logged on pflog0 exactly every 10 seconds), but it looks like > nothing can be sent.A bridge on a stick only works under certian conditions, mainly support by the switch its connected to. Its obviously working for you for a period of time so thats probably ok. Make sure that you dont use scrubbing with pf as the bridge doesn't handle the fragment reassembly at the moment. Can you also try a few things when it stops - flush the bridge table using ifconfig - use tcpdump on the bridge interface to see if its forwarding packets cheers, Andrew
Yuriy Tsibizov
2006-Apr-20 11:29 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
> On Wed, Apr 19, 2006 at 04:15:10PM +0400, Yuriy Tsibizov wrote: > > I run a 6.1-RC (as of Apr, 14th) and after a day nothing > can be sent from em interface. > > > > It is a filtering+accounting bridge-on-a-stick, with em interface > > connected to 3Com Switch 4200 100Mbit (not Gbit) port with 4 tagged > > vlans on it: > > > > Polling was not used (I'll try to work with polling enabled on em0). > > > > ALTQ is not configured, pf has all block rules logged and I > can't see anything relevant logged to pflog. > > > > Traffic is received (I can see pairs of OSPF HELO messages from ISP > > router logged on pflog0 exactly every 10 seconds), but it looks like > > nothing can be sent. > > A bridge on a stick only works under certian conditions, > mainly support > by the switch its connected to. Its obviously working for you for a > period of time so thats probably ok. > > Make sure that you dont use scrubbing with pf as the bridge doesn't > handle the fragment reassembly at the moment.I had it, and will try to work without it. Thank you for your help, Yuriy.
Yuriy Tsibizov
2006-Apr-20 12:21 UTC
em (+pf+vlan+bridge+netgraph) stop sending packets (?)
> > On Wed, Apr 19, 2006 at 04:15:10PM +0400, Yuriy Tsibizov wrote: > > > I run a 6.1-RC (as of Apr, 14th) and after a day nothing > > can be sent from em interface. > > > > > > It is a filtering+accounting bridge-on-a-stick, with em interface > > > connected to 3Com Switch 4200 100Mbit (not Gbit) port > with 4 tagged > > > vlans on it: > > > > > > Polling was not used (I'll try to work with polling > enabled on em0). > > > > > > ALTQ is not configured, pf has all block rules logged and I > > can't see anything relevant logged to pflog. > > > > > > Traffic is received (I can see pairs of OSPF HELO > messages from ISP > > > router logged on pflog0 exactly every 10 seconds), but it > looks like > > > nothing can be sent. > > > > A bridge on a stick only works under certian conditions, > > mainly support > > by the switch its connected to. Its obviously working for you for a > > period of time so thats probably ok. > > > > Make sure that you dont use scrubbing with pf as the bridge doesn't > > handle the fragment reassembly at the moment. > I had it, and will try to work without it.One question - if pf was the cause, should it recover when I disable pf? As I remeber, the box was still silent on vlan interfaces, when pf was disabled with "/etc/rc.d/pf stop". Yuriy