The default value for this option (as we did not explicitly set it we are using the default) is 0/none which means all slave interfaces will use the same MAC address. What you need to know is that there is another parameter/option called fail_over_mac. In our cat ifcfg-bond0īONDING_OPTS="mode=active-backup miimon=100" But both slaves had the same mac address, how that? It all depends on the bonding options. This was not so scary as we knew that the bonding interface will get one of the mac addresses of the slave interfaces. In addition to that the bonding interface bond0 has the same mac address as well:
Inet6 fe80::a00:27ff:fe85:2d8/64 scope link tentativeĪs you can see the interfaces enp0s8 and enp0s10 have exactly the same MAC address. Link/ether 08:00:27:85:02:d8 brd ff:ff:ff:ff:ff:ffĥ: enp0s10: mtu 1500 qdisc pfifo_fast master bond0 state UP qlen 1000Ħ: bond0: mtu 1500 qdisc noqueue state UP qlen 1000 Inet6 fe80::a00:27ff:fe5c:b0e5/64 scope linkĤ: enp0s9: mtu 1500 qdisc pfifo_fast master bond0 state UP qlen 1000 Valid_lft 86295sec preferred_lft 86295sec The situation we faced looked like ip aġ: lo: mtu 65536 qdisc noqueue state UNKNOWN qlen 1 For the little demo for this post I’ll use CentOS 7.4 but this should not really matter. In the end it is clear and not strange anymore but for a moment we really thought we are hitting a bug in the operating system, which was RedHat 7.4. Yes, sounds strange, but exactly this is what we faced today.
0 kommentar(er)
