I have been facing issues with this model for almost a year. For the issue of hanging ports, Mikrotik support suggested a temporary solution using a scheduled script:
At some point, I received a pre-release version of RouterOS, which seemed to resolve the port hanging issue. Later, an official stable firmware update was released on July 16, 2024, and I received a notification from Mikrotik confirming that the solution for the port hang issue was included in this update. I updated to the latest stable release and have been keeping it up to date since then.
However, on RouterOS 7.15.3, the port hanging problem reappeared, so I re-enabled the scheduled script. Since then, I have observed the script resetting all ports only once. I am uncertain whether this was due to the port hanging issue or simply a lack of ping response during other network activities at the time. I am continuing to monitor the situation closely.
However, another problem occurs periodically: sometimes the network slows down almost to a complete stop. After extensive testing and debugging, I discovered that the switch occasionally sends random data to all network ports as packets. For example, I found a packet made from a fragment of JSON data, and the source (SRC) and destination (DST) MAC addresses were random (not present in the network). These packets are transmitted across the entire network as the switches attempts to locate the DST MAC address (receiver) but are ultimately dropped at the endpoints.
Because the SRC MAC address is also random, it gets registered in the MAC tables of all network devices. Due to the high volume of random packets, the MAC tables on all devices grow rapidly, clearly indicating that the CRS354-48P-4S+2Q+ switch has started malfunctioning.
Throughout this time, I performed various reconfigurations and tests. Today, the switch malfunctioned again. My test setup consisted of only a router and the switch, with all other ports disabled except one connected to the router. The router was configured with only a WAN interface and one port connected to the switch. The switch was transmitting random packets to the router, but the traffic from the router was clean. After rebooting the switch, the random traffic stopped in both directions. The CPU usage of the switch also dropped from ~30% to 6–9%. RouterOS v7.16.2, firmware updated.
It seems like there is some kind of buffer overflow in the OS or a memory issue.
I have provided all this information to Mikrotik support and am currently waiting for their response. In the meantime, I am preparing the device for return, either to replace it with another model or switch to a different brand, as I no longer have the time or energy to deal with these issues in a production network.
Code:
if ([/ping gateways_IP count=10] = 0) \do={ /interface ethernet disable [find where switch=switch1]; :delay 3s; /interface ethernet enable [find where switch=switch1]; log warning "Resetting switch1 !!!"}However, on RouterOS 7.15.3, the port hanging problem reappeared, so I re-enabled the scheduled script. Since then, I have observed the script resetting all ports only once. I am uncertain whether this was due to the port hanging issue or simply a lack of ping response during other network activities at the time. I am continuing to monitor the situation closely.
However, another problem occurs periodically: sometimes the network slows down almost to a complete stop. After extensive testing and debugging, I discovered that the switch occasionally sends random data to all network ports as packets. For example, I found a packet made from a fragment of JSON data, and the source (SRC) and destination (DST) MAC addresses were random (not present in the network). These packets are transmitted across the entire network as the switches attempts to locate the DST MAC address (receiver) but are ultimately dropped at the endpoints.
Because the SRC MAC address is also random, it gets registered in the MAC tables of all network devices. Due to the high volume of random packets, the MAC tables on all devices grow rapidly, clearly indicating that the CRS354-48P-4S+2Q+ switch has started malfunctioning.
Throughout this time, I performed various reconfigurations and tests. Today, the switch malfunctioned again. My test setup consisted of only a router and the switch, with all other ports disabled except one connected to the router. The router was configured with only a WAN interface and one port connected to the switch. The switch was transmitting random packets to the router, but the traffic from the router was clean. After rebooting the switch, the random traffic stopped in both directions. The CPU usage of the switch also dropped from ~30% to 6–9%. RouterOS v7.16.2, firmware updated.
It seems like there is some kind of buffer overflow in the OS or a memory issue.
I have provided all this information to Mikrotik support and am currently waiting for their response. In the meantime, I am preparing the device for return, either to replace it with another model or switch to a different brand, as I no longer have the time or energy to deal with these issues in a production network.
Statistics: Posted by nezmogus — Wed Jan 01, 2025 3:07 pm