After upgrading from v4.17 to v5.6 we noticed that the pcq’s were generating at times large numbers of subqueques. The screen shot below shows 418 subqueues in the queue ″Upload / Priority-Low-Up″. There were only a few active clients connected at the time (<10). Excessive number of subqueues only occurs in the last 2 queues. About half the time the number of subqueues is small and within the expected range. But about 30 % of the time the number of subqueues is between 40 and 200 and occasionally up to 600. The queue tree is set up such that each client should get only 1 subqueue, thereby sharing the bandwidth equally. Below are also the screen shots of the queue types ″priority-up″ and ″priority-down″.
We also upgraded a network from v4.17 to v5.6 with 100+ clients and heavy traffic. There, the upload was completely overloaded and thereby also blocking the download. We quickly had to go back to v4.17.
Here is further information to the previous post “v5.6, pcq generates excessive number of subqueues”.
I downgraded to v5.0rc1 and then upgraded again to v5.0rc7. It turns out that the problem described in the previous post was not present in v5.0rc1, but was present in v5.0rc7 and behaves like in v5.6. The 2 screen shots correspond to the 2 versions. There were only 3 clients connected and it cannot be that they generate 102 subqueues. The 2 screen shots were taken 20 min apart.
This problem has been present since at least v5.0rc7. Has nobody else observed this?
After further investigation I have now found that the problem with excessive number of PCQs only occurs if under Bridge / Settings the square “Use IP Firewall For PPPoE” is ticked.
