GermanRed+
Active member
I can confirm that "addnode" to another MN will create the high CPU load and high memory usage problem. Eventually, the MN will be kicked off with this high CPU load and memory consumption. Something is still wrong in v0.11.2.17.
EDIT: What I did is "addnode another_MN_IP_address_on_the_same_ISP:9999 add". Then, wait until the two MNs are connected as peers. I use "lsof -i -n" to check that or you can use "getpeerinfo" instead. The time to get these two nodes connected can take some time. When they are connected, you will see the CPU load increases and memory usage starting to rise. I just tested it on a node that ran without being kicked off for 15 hours. Right after the MNs are connected as peers, the CPU load started to ramp up and the memory usage gradually increased.
I haven't tried this on MN hosted by another ISP as I do not want to kick someone off. Could someone with MNs on two different ISPs try this? You can restart the nodes once you see CPU load over 100% and memory usage more than 366 MB. That will confirm the problem without being kicked off because a simple restart of the nodes will prevent that.
EDIT: What I did is "addnode another_MN_IP_address_on_the_same_ISP:9999 add". Then, wait until the two MNs are connected as peers. I use "lsof -i -n" to check that or you can use "getpeerinfo" instead. The time to get these two nodes connected can take some time. When they are connected, you will see the CPU load increases and memory usage starting to rise. I just tested it on a node that ran without being kicked off for 15 hours. Right after the MNs are connected as peers, the CPU load started to ramp up and the memory usage gradually increased.
I haven't tried this on MN hosted by another ISP as I do not want to kick someone off. Could someone with MNs on two different ISPs try this? You can restart the nodes once you see CPU load over 100% and memory usage more than 366 MB. That will confirm the problem without being kicked off because a simple restart of the nodes will prevent that.
Last edited by a moderator: