How to clear the ARP cache on Linux?
There are several reasons when you might need to clear your ARP cache. There are two common ways on Linux systems, typically using the arp or ip utility. Depending on your Linux distribution and the availability, we suggest using the ip tool.
Clearing cache with ip
Newer Linux distributions have the ip utility. The ip tool has a more advanced way to clear out the full ARP cache.
ip -s -s neigh flush all
The first -s will provide a more verbose output. By adding one more, we can select the neighbor table. The neighbor table with the ip command equals both the ARP and NDISC cache. Note that the -s options are not available on all versions of the ip command. If it not supported for your version of ip, then simply remove them from the command.
The output of the flush all command will produce the following output.
The ARP cache is cleared, with verbose output
Clearing cache with arp command
The arp utility does not accept an option to clear the full cache. Instead, it allows to flush out entries found with the -d option.
arp -d 192.168.1.1
After deleting, have a look with the arp utility again to see the new list:
arp -n
The output of this command will typically show the active ARP entries.
Address HWtype HWaddress Flags Mask Iface 192.168.1.1 (incomplete) eth0 192.168.1.2 ether 00:02:9b:a2:d3:f3 C eth0 192.168.1.3 ether 00:02:9b:d9:d1:a2 C eth0
The 192.168.1.1 entry now shows as incomplete, which means the ARP entry will be refreshed when it is needed again.
Conclusion
Depending on your distribution, the ip utility is quicker if you want to flush out the full ARP cache. For individual entries, the arp tool will do the job as quickly. Both tools are available for most distributions, including Arch Linux, CentOS, Debian, Fedora, RHEL, and Ubuntu.
Did this article to clear the ARP cache help you as well? Wonderful! Become part of the community and share this on social media to let others know. Got questions or suggestions? Join us in the comments.
Security scanning with Lynis and Lynis Enterprise
4:05 AM
I need to apply flush-clean the arp table, and to have one option as Winsock for the case of win7 and if exist clean the register as Ccleaner, because I have the problem ;
ubuntu@ubuntu:~$ netstat -nat
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address Foreign Address State
tcp 0 0 127.0.0.1:631 0.0.0.0:* LISTEN
tcp6 0 0 ::1:631 :::* LISTEN
tcp6 1 0 ::1:53003 ::1:631 CLOSE_WAIT
How can to resolve this case , in other case appear following: I think that need to clean , by the “listen”
Proto Recv-Q Send-Q Local Address Foreign Address State
tcp 0 0 127.0.1.1:53 0.0.0.0:* LISTEN
tcp 0 0 127.0.0.1:631 0.0.0.0:* LISTEN
tcp 0 0 172.252.37.7:46799 94.31.29.192:80 ESTABLISHED
tcp 0 0 172.252.37.7:51130 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:45273 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:44964 149.210.134.182:443 ESTABLISHED
tcp 0 0 172.252.37.7:45270 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:60107 64.233.176.95:80 TIME_WAIT
tcp 0 0 172.252.37.7:35899 216.58.219.110:80 TIME_WAIT
tcp 0 78 172.252.37.7:40483 216.58.219.72:443 LAST_ACK
tcp 0 0 172.252.37.7:59036 216.58.219.67:80 TIME_WAIT
tcp 0 78 172.252.37.7:38157 64.233.185.94:443 LAST_ACK
tcp 0 0 172.252.37.7:45275 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:51127 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:45271 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:41876 216.58.219.68:80 ESTABLISHED
tcp 0 0 172.252.37.7:43539 64.233.185.154:80 ESTABLISHED
tcp 0 0 172.252.37.7:45272 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:39388 24.139.135.147:80 ESTABLISHED
tcp 0 0 172.252.37.7:60106 64.233.176.95:80 TIME_WAIT
tcp 0 0 172.252.37.7:51131 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:37724 216.58.219.78:80 TIME_WAIT
tcp 0 0 172.252.37.7:59708 216.58.219.66:80 ESTABLISHED
tcp 0 0 172.252.37.7:51586 173.194.219.95:80 TIME_WAIT
tcp 0 0 172.252.37.7:51128 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:37227 216.58.219.98:80 TIME_WAIT
tcp 0 0 172.252.37.7:45274 91.189.95.69:80 ESTABLISHED
tcp 0 0 172.252.37.7:51587 173.194.219.95:80 TIME_WAIT
tcp 0 0 172.252.37.7:35809 64.233.176.94:80 ESTABLISHED
tcp 0 0 172.252.37.7:59035 216.58.219.67:80 TIME_WAIT
tcp 0 0 172.252.37.7:51132 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:51129 91.189.94.232:443 ESTABLISHED
tcp 0 0 172.252.37.7:35666 216.58.219.110:80 TIME_WAIT
tcp 0 0 172.252.37.7:34426 173.194.219.94:80 ESTABLISHED
tcp6 0 0 ::1:631 :::* LISTEN
tcp6 1 0 ::1:53003 ::1:631 CLOSE_WAIT
Thanks for your attention ,
9:27 AM
These are your active connections (to your web server software). ARP is a protocol one level below these network connections. For details for flushing the ARP table, see the article on how to do that. For easily resetting the active connections, reload your web server daemon (Apache, nginx etc). The ports which state “LISTEN”, have a daemon running (53 = DNS, 631 = SAMBA or CUPS).
11:16 AM
The statement “The second one defines the neighbor table” is not accurate. Both -s are for verbosity (providing 2 increases it). The basic command is simply “ip neigh flush all”
2:47 PM
Depending on your version of the ip utility, you may not have the -s option available. The text have been extended to reflect that. Thanks for the feedback!
8:36 AM
ip neigh flush all does not flush the cache completely, i.e. it does not delete the entries from the neighbor table. Instead it only clears the cached MAC addresses in the neighbor table, i.e. it sets all entries to state FAILED. But the entries with IP and IPv6 address are still kept in the cache. In former times the kernel expired entries after some time without usage, but unfortunately, this is not done anymore and there seems to be no way to remove entries manually.
Even a normal user could fill the neighbor cache with lots of entries which will stay until the next reboot. E.g. with “for n in {1..255}; do ping -c1 10.0.0.$n; done”
9:15 PM
I expected arp entries of things in use to be more or less immediately re-populated, as the next packet to them would initiate an arp query and response in a few milliseconds. This was not the case. It took about 2 minutes for everything to return to normal. So be advised.
11:03 AM
Thanks for sharing your insight!