Hello
In two Mellanox SN2410 switches configured with MLAG, we have seen a large number of packets dropped from the bonding of the hosts connected to the switches. We have read that it is a BUM protection method but we need to know if it is possible to adjust the values given that in one day we have 1,000,000 packets lost on a bond
Regards
There are a lot of questions from this topic.
What OS is running?
What version?
What counters are identifying your lost packets on what interfaces?
Hello
The OS is cumulus version 5.9 (Spine Switches) and 5.5 in Leaf Switches.
The problem is the high number of dropped packages, in the LACPs that go from Spine to Leaf and vice versa.
The counters have been cleaned, and in 1 day they can reach 2,000,000 dropped packages in RX_DROP and TX_DROP
Support has indicated that this is normal and that it is a loop protection system that can be caused by high multicast traffic.
We have carried out LABS with other vendors (Cisco - Huawei) and they do not show this behavior, and even the CPU usage has exceeded 90%