87

Since ifconfig is apparently being deprecated in major Linux distributions, I thought I'd learn something about the ip tool that's supposed to be used instead of ifconfig.

And here I ran into a problem: when run on its own, ifconfig shows the number of bytes received/transmitted on each interface besides other info. I couldn't find a way to get this from ip. Is there no such function in this tool? What other built-in tools could I use for getting those stats?

justinas
  • 1,008
  • 1
  • 8
  • 8

5 Answers5

79

The ip command which is part of the the iproute2 package is the new tool. The link subcommand is for managing the devices/interfaces.

If you can get the stats of an interface using ip -s link

root:~# ip -s link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    RX: bytes  packets  errors  dropped overrun mcast
    50679705   529967   0       0       0       0
    TX: bytes  packets  errors  dropped carrier collsns
    50679705   529967   0       0       0       0
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
    link/ether 00:1d:7d:aa:e3:4e brd ff:ff:ff:ff:ff:ff
    RX: bytes  packets  errors  dropped overrun mcast
    187663757  308710386 0       0       0       0
    TX: bytes  packets  errors  dropped carrier collsns
    4051284587 532435117 0       0       0       0
Zoredache
  • 130,897
  • 41
  • 276
  • 420
  • 9
    While the `proc` answer is clean and lightweight, this answer is probably the closest to a "replacement" for the `ifconfig` tool because it was explicitly designed with replacement of `ifconfig` in mind. – Moshe Katz Aug 29 '13 at 20:40
  • 8
    `ip -s link show dev eth0` to show only traffic on eth0 – pim May 09 '18 at 11:33
  • 9
    To get this in human readable form (reporting in mb/gb/tb rather than bytes), use the `-h` flag. For instance, `ip -h -s link`. – Joseph Redfern Apr 25 '19 at 10:45
69

Another option is to use the /proc filesystem. The /proc/net/dev file contains statistics about the configured network interfaces. Each line is dedicated to one network interface and it contains statistics for receive and transmit. The statistics include metrics such total number of received/transmittted bytes, packets, drops, errors and so on.

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:    29846937   129576     0    0    0     0          0       0 29846937   129576     0    0    0     0       0          0
 wlan0:    9467393340 8027251    0    0    0     0          0       0 2559312961 5896509    0    0    0     0       0          0

Or you can try the netstat command which can display all network interfaces and related statistics:

netstat -i

Iface   MTU Met    RX-OK RX-ERR RX-DRP RX-OVR    TX-OK TX-ERR TX-DRP TX-OVR Flg
lo         65536   0   130435      0      0      0   130435      0      0      0 LRU
wlan0       1492   0  8028018      0      0      0  5897361      0      0      0 BMRU
dsmsk80
  • 5,817
  • 18
  • 22
  • `/proc/net/dev` has the cleanest format of all of the solutions IMO, accepting. – justinas Aug 25 '13 at 15:23
  • 1
    Stats showed by `cat /proc/net/dev` are for how long time ? – lalebarde Jul 18 '14 at 10:35
  • How do you get tx/rx from `/proc/net/dev` ? – Piotr Kula Jun 08 '15 at 21:31
  • 1
    Try to parse 3. and 11. field of the output, something like awk '/:/ { print($1,$3, $11) }' < /proc/net/dev – dsmsk80 Jun 09 '15 at 13:04
  • netstat is deprecated along with ifconfig. On Linux, not sure about other *IX varietals. – Mike S Apr 30 '21 at 18:47
  • `netstat -i` cuts off interface names after the 8th char, not very usable if your system has longer interface names (which mine does). Recommend using `ss -i` instead. `ss` is the replacement to netstat. Unrelated, many flags that were used in `netstat` are supported in `ss` – Dave Aug 09 '22 at 17:39
62

You can get all necessary info via proc

# cat /sys/class/net/eth0/statistics/rx_bytes
# cat /sys/class/net/eth0/statistics/rx_packets

# cat /sys/class/net/eth0/statistics/tx_packets
# cat /sys/class/net/eth0/statistics/tx_bytes

Also you can use iptables and parse output.

For received packets 

# iptables -L INPUT -n -v

for transmitted packets

# iptables -L OUTPUT -n -v

If server is a gateway, then you should also parse FORWARD chain

ALex_hha
  • 7,193
  • 1
  • 25
  • 40
  • This answer is the most simple, I was able to build a script with this fairly easily. Also I use `mmwatch` (https://github.com/cloudflare/cloudflare-blog/blob/master/2017-06-29-ssdp/mmwatch) to turn the values in to rates. However, do not use `iptables` now, it is deprecated. – Dave Aug 09 '22 at 18:04
  • to have a single command to use when checking on servers I use `grep . /sys/class/net/$(ip route show|awk '/default/{print $5}'|head -1)/statistics/*|grep -v ':0$'` or - if you know the name of the nic - `grep . /sys/class/net/wlp0s20f3/statistics/*|grep -v ':0$'` – lpaseen Nov 09 '22 at 19:18
1

You can read the file /sys/class/net/wlp3s0/statistics/rx_bytes and get the rx_byes directly without calling another command, vnstat is also good. Linux stores all information in files as I know, so better to find those files and get information. Finding the relevant file is the challenge.

Dinuka Salwathura
  • 111
  • 6
  • 2
    Welcome to the community! Well done, that's correct, but note that some old kernels don't provide these files. Wish you luck, and success! ;) –  Mar 25 '17 at 17:56
0

You can also use ethtool:

ethtool -S eth0

The benefit to ethtool, is you can see each rx / tx queue and how they are balanced.

I found watch is a great tool to update the screen using all these tools. However most of these tools display traffic as bytes or packets per second. I find it difficult to convert bytes to gigs or pps to millions of packets persecond on the fly. Just a tip, check out Cloudflare's mmwatch tool. It does the same as watch, but will do the calculations automatically and display the output to the screen like watch. (Example: So you will see 1.2m/s for 1.2 million packets per-second)

https://github.com/cloudflare/cloudflare-blog/blob/master/2017-06-29-ssdp/mmwatch

Also, ethtool -S eth0 gives A lot of output, I grep the following way to show receive queues and discards also with mmwatch:

mmwatch "ethtool -S eth0 | grep 'rx-\|discard'"
Dave
  • 229
  • 2
  • 10