I am experiencing frequent state changes of eth0 in the kernel log, the messages look like this:
[514004.065607] bcmgenet fd580000.ethernet eth0: Link is Down
[514007.137676] bcmgenet fd580000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
During the course of the last 24 hours I got this down/up combination about 900 times. The flapping started to appear more frequently when the load of the machine increased. The load is moderate with a load average of 1,47, 1,43, 1,27, a consistent network use of between 60 and 80 Mbit/s, and one of four CPUs at 100%.
The "down" periods seem to be sufficiently short to not cause any noticable connection problems when using TCP, but I see packet loss of approx. 10% for ICMP requests. Also, the RTT to the interceptor in the local network varies a lot, it should always be <1 ms, but min/avg/max/mdev in fact is 0.288/4.601/9.523/3.217 ms as measured over 300 requests using ping. RTT to a regular RPi4 connected to the same switch is 0.4 ms on average.
I am using kernel 5.15.32-v8 with the OS as provided by Axzez.
Has anyone else experienced anything similar?
I've continued testing and it seems that the combination CM4/Interceptor is struggling with low to intermediate load on network and SATA (se my previous post six days ago).
I'd like to know whether I can do something about this, or whether I need to search for an alternative hardware platform to build a NAS for a SOHO setting.
Any suggestions?