Max Queue Errors

[Steve G]

May 15 09:07:29 bwmgr kernel: Queue Full: len:3000 max:3000

We're getting these messages in our log. What does it mean and how can we fix it?

[Bob]

May 15 09:07:29 bwmgr kernel: Queue Full: len:3000 max:3000

We're getting these messages in our log. What does it mean and how can we fix it?

You need to increase your max_qlen setting:

sysctl -a | grep max_qlen

The default is 3000. This is set when you create the rule, so you need to set this in your bootup sysctl config:

edit /etc/sysctl.conf:

bwmgr.max_qlen=5000

And then reboot. If you’re just having a problem with 1 rule you could just do

sysctl bwmgr.max_qlen=5000

and then re-create the rule

[Dennis]

Note that this will be dynamic shortly. The queue depth should be calculated based on the allowed bandwidth. In fact I’m going to add that as a priority task.

TCP traffic gets window shaped so if you’re blowing. your queue it means you have a lot of icmp or udp traffic and you might want to see what they’re doing if this is an individual customer rule.

[Bob]

As of 8.071 you can override the default max queue length (ie the number of packets that can be delayed) in both profiles and individual rules

[Dennis]

The max_qlen setting for each rule is now dynamic based on the bandwidth setting for the rule. It will not exceed the sysctl setting for bwmgr.max_qlen. You can override this in profiles or individual rules