CentOS

It seems that vlan interfaces over bond doesn't work correctly. Specifically traffic statistics

I have the following network config

[code]
# cat /etc/sysconfig/network-scripts/ifcfg-bond1
DEVICE=bond1
ONBOOT=yes
BOOTPROTO=none

# cat /etc/sysconfig/network-scripts/ifcfg-bond1.210
# Intel Corporation 82576 Gigabit Network Connection
DEVICE=bond1.210
BOOTPROTO=none
IPADDR=192.168.210.1
NETMASK=255.255.255.0
ONBOOT=yes
VLAN=yes

# cat /etc/sysconfig/network-scripts/ifcfg-eth2
# Intel Corporation 82576 Gigabit Network Connection
DEVICE=eth2
BOOTPROTO=none
ONBOOT=yes
HWADDR=00:1B:21:63:55:44
MASTER=bond1
SLAVE=yes

# cat /etc/sysconfig/network-scripts/ifcfg-eth3
# Intel Corporation 82576 Gigabit Network Connection
DEVICE=eth3
BOOTPROTO=none
ONBOOT=yes
HWADDR=00:1B:21:63:55:45
MASTER=bond1
SLAVE=yes

# cat /proc/net/bonding/bond1
Ethernet Channel Bonding Driver: v3.4.0-1 (October 7, 2008)

Bonding Mode: load balancing (xor)
Transmit Hash Policy: layer2 (0)
MII Status: up
MII Polling Interval (ms): 80
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth2
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 00:1b:21:63:55:44

Slave Interface: eth3
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 0
Permanent HW addr: 00:1b:21:63:55:45
[/code]

But when I try to get statistics of the interface I can't see any Incoming data
[code]
# cat /proc/net/dev | grep bond1.210
bond1.210: 0 0 0 0 0 0 0 0 101106032062 446733933 0 0 0 0 0 0
[/code]

At the same time on eth2/eth3 all shows correctly
[code]
# cat /proc/net/dev | grep eth2
eth2:2401029223576 2945035953 0 0 0 0 0 262013 1753706319816 2729232623 0 0 0 0 0 0

# cat /proc/net/dev | grep eth3
eth3:1342749003523 1755079950 0 0 0 0 0 365234 1620089577678 1939377201 0 0 0 0 0 0
[/code]

I ran into this problem when I try to draw pps statistic (packet per second) via zabbix

On CentOS-6.2 all works fine

[code]
# cat /proc/net/dev
Inter-| Receive | Transmit
face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
lo: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
eth0: 23152 224 0 0 0 0 0 0 21893 171 0 0 0 0 0 0
eth1: 4308 44 0 0 0 0 0 0 21303 196 0 0 0 0 164 0
bond0: 27460 268 0 0 0 0 0 0 43196 367 0 0 0 0 164 0
bond0.212: 4136 50 0 0 0 0 0 0 5116 56 0 0 0 0 0 0
[/code]

Аny suggestion?

Since everything in /proc comes from the kernel or from kernel modules, anything that's missing must be because of a bug in the kernel. You can either report the problem on the upstream bugzilla and hope that they deem it important enough to fix (I have my doubts) or you can use a different kernel. The ELRepo has a kernel-ml repo which is disabled by default and that contains a recompiled mainline kernel that you could try but be aware that it takes you outside the realms of CentOS support.

[quote]you could try but be aware that it takes you outside the realms of CentOS support.[/quote]
I am not so brave to test such kernel in the production environment ;) But I will try on the test box

[quote]You can either report the problem on the upstream bugzilla and hope that they deem it important enough to fix (I have my doubts) [/quote]
Can I create report about this problem even if I use CentOS, not RHEL? (I know that CentOS binary compatible with RHEL). If I can, what should I specify in system description - CentOS or RHEL? :)

Upstream Bugzilla has an entry for CentOS under External Bugs. It would be best to first report it for CentOS and let the CentOS developers decide if it needs to go upstream. If it does they may either make the report, or ask you to do so.

Thanks, but it seems there is no need. See https://www.centos.org/modules/newbb/viewtopic.php?topic_id=34926&forum=40

OK - so does that mean both threads can be marked [SOLVED]?

I think so, but unfortunately I can't say what exactly cause issue with Incoming statistics

If I can tomorrow I will try to reproduce this issue on the test box with RHEL-5.7. Anyway I will inform you.

Standing by...