We were questioned by the operating systems team, which reported to us that on some servers were with the error below in dmesg,
"Oct 5 03:46:15 srvpwcpvlbr02 kernel: [29989.079976] oneagentnetwork: page allocation failure. order:4, mode:0xd0"
According to them, the normal is the application allocating memory pages with 4 KB sizes, the agent in the case is attempting to allocate 64 Kb.
Does the agent really need to allocate this amount of memory? Do you have any way to optimize this?
Solved! Go to Solution.
This question might be hard to get answered on the forum. You might be better off creating a support case which can be formally escalated to engineering where this type of design question can be properly answered.
Hey i made a chat with support a few weeks ago for the same topic:
(In our case: Sep 29 08:06:54 XXXXXXXXXXXX kernel: oneagentnetwork: page allocation failure. order:4, mode:0xd0 )
This is what i got:
This message does not indicate any problem with Dynatrace components. The messages are caused by the fact, that libpcap (third party library we use to capture network traffic) is unable to find continuous block in kernel address space. If the continuous address space may not be found, libpcap will use ioctl to pass the captured traffic to Network Agent instead. This will affect the Network Agent performance to some degree, but should not have impact on any other system components.This is a known message in Dynatrace that doens't represent a problem in either your environment or in Dynatrace components. As of right now there is no way to "stop" it from show up at the messeges log.