Cache-out traffic decreased - Troubleshooting Article

Overview

After restarting your PeerApp system, you might observe a noticeable decrease in the cache-out performance. 

Workflow

Instructions

Check the System Performance

Check the system performance and look for any unusual processes or issues. Observe if there is a buffer pool-related issue. If there is, calibrate it with a system restart. 

This can be done in the UBView by checking;

Home | Statistics > Interfaces

Home | Statistics > Downstream Bandwidth > Cache Out 

and looking for any gaps or unusual behaviors in the graphs.

Check the Statistics

Check the statistics of the system on and around the day when the problem started. See if there are any gaps or unusual changes in the traffic statistics. If there is a sudden change in the statistics, find out what happened or was done with the system on that day and time to get to the root cause.

In the UBView, check;

Home | Statistics > Downstream Bandwidth > Total Downstream

Home | Statistics > Downstream Bandwidth > Cache Productivity 

to get detailed statistics.

Observe the Traffic Pattern

Observe the traffic pattern and look for any sudden and unusual changes. Note that PeerApp supports UDP traffic but due to the fact of UDP protocol, the caching system can produce more traffic for TCP traffic. Non-cachable traffic is huge in the UDP traffic therefore cache out traffic is less with UDP traffic. If that is the case and UDP traffic has been allowed, try disabling that and passing maximum TCP traffic which should increase cache-out traffic and resolve the issue.

This can be done by checking 

Home | Statistics > Downstream Bandwidth > Total Downstream

and looking for any unusual changes in the traffic pattern especially around the day the issues started to appear. 

If none of the above steps helped, parameter tuning might be needed to solve the problem. Raise a support ticket and PeerApp support will be happy to help resolve your issue.

<supportagent>

If the network protocol change isn't the cause and the cache-out traffic went down due to another reason e.g. restarting the system, it is most likely that after a system restart, the TCP parameters were not set correctly. A parameter fine-tuning in a cluster configuration would most probably resolve this issue.

</supportagent>