[rsyslog-notify] Forum Thread: Re: Dequeue Perfomance - (Mode 'reply')

[noreply at adiscon.com]

User: kamermans 
Forumlink: http://kb.monitorware.com/viewtopic.php?p=25546#p25546

Message: 
----------
Update: there is a problem with the route from my ISP (Verizon FiOS) and
Amazon EC2 us-east-1 (which, ironically, is 5 miles from my house), which
is causing packet loss on the order of 10-20% each night between
8:00PM-11:30PM EST very predictably.  I have confirmed it in two separate
locations in my region.  I'm putting this on hold as my logs clearly
indicate a correlation between packet loss and rsyslog queue size, and the
packet loss always happens first.  The dequeue is slow because the loss
happens for a few hours each night.  When it clears up, the dequeue floods
data back to the receiver and all is well (assuming I don't OOM first).

I'm putting this on hold for now, but I really think there might be a
memory management problem somewhere, as rsyslog seems to eat way more RAM
than just the queues would consume during times of intermittent
connectivity.  You can see from my memory graphs that the memory
consumption is almost perfectly linear until it crashes - I have multiple
events logged like this.

Thanks for all your help troubleshooting this!