First I am going to define OLSR storm because that is not an official term, but it is being used here in Southern California. What we call an OLSR storm is a condition where OLSR packets are being distributed in such a manner that it functionally crashes the mesh network. The symptoms include that nodes can not be seen, or link qualities plummet network wide. This specifically includes DtD links and tunnels - you know there is a problem when a DtD link shows a LQ of 10%! Usually eventually the storm will subside by itself, but that can take several hours. Manually breaking links (usually tunnels because that is easy to do) helps the process along. We have several people here in Southern California that have set up various detection methods to automatically detect OLSR storms, and in the case of one person, he can usually detect the node at the center of the storm.
With that all said, there seems to be a possible pattern to these. Node restarts. In many of the cases of OLSR storms where a node can be identified as the center of the storm, frequently that node had just been restarted or initially powered up - determined by uptime. Last evening one of my nodes was identified as the center of an OLSR storm. I initially noticed the problem when I could not maintain a SSH connection for more than a few seconds into a node that has a DtD link to the node I was connected to. Just moments before, I had updated that storm center node from firmware 3.20.3.1 to Nightly Build 1693. This particular node is a Mikrotik hAP This pattern has been seen before.
Is there a pattern? Maybe, maybe not. However, we in SoCal have been looking for something common for quite a while. So I ask you all, have you seen this problem, and if yes, anything that would go along with this as a pattern? Or something else?
With that all said, there seems to be a possible pattern to these. Node restarts. In many of the cases of OLSR storms where a node can be identified as the center of the storm, frequently that node had just been restarted or initially powered up - determined by uptime. Last evening one of my nodes was identified as the center of an OLSR storm. I initially noticed the problem when I could not maintain a SSH connection for more than a few seconds into a node that has a DtD link to the node I was connected to. Just moments before, I had updated that storm center node from firmware 3.20.3.1 to Nightly Build 1693. This particular node is a Mikrotik hAP This pattern has been seen before.
Is there a pattern? Maybe, maybe not. However, we in SoCal have been looking for something common for quite a while. So I ask you all, have you seen this problem, and if yes, anything that would go along with this as a pattern? Or something else?