Solutions

Various ProxySG high CPU causes and related solutions

Solutions ID:    KB3894
Version:    9.0
Status:    Published
Published date:    07/14/2010
Updated:    06/30/2011
 

Problem Description

High CPU utilization is being recorded on the ProxySG
Need assistance troubleshooting high CPU utilization on the proxy
What is causing high CPU utilization on my ProxySG?
The ProxySG is running at 100 percent, and causing slow response times for users.
Messages you may see in your event log.
Health Monitor (WARNING): CPU Utilization above 80 percent for last 120 seconds"  64 700101:1 Mailed ../notifier.cpp:149
Health Monitor (CRITICAL): CPU Utilization above 95 percent for last 120 seconds"  64 700102:1 Mailed ../notifier.cpp:149

Resolution

There are many different reasons why high CPU utilization may occur on the ProxySG platform.  This document attempts to be a clearing house for the various reasons why high CPU utilization may occur on the ProxySG platform.  There may be other reasons not documented here.  Over time, as new sources of high CPU utilization occur, they will be added to this document.

 

Basic troubleshooting steps for high utilization issues:

It is helpful to be aware of external, use behavior factors, that may cause excessive load on the proxy. 

Questions you should be answering here are:

  • Is the proxy continuously at high CPU, or does it peak at certain times?
  • Does CPU utilization come down after production hours?
  • Is CPU utilization peaking periodically then going back to “normal”?  If so, can you graph out the peaks, and identify a trend?

Once you have identified which proxy it is, when, and for how long the high CPU event occurs, we then need to get to the root cause of why it occurs.  To do so, we need to establish which SGOS service, or services, are putting the most burden on the CPU.  We do this by enabling the CPU monitor, and then taking snapshots at five minute intervals.  By watching the trend on each CPU monitor statistics, using each successive snapshots , we can then draw a conclusion as to which service is taking the most CPU cycles.

To do this, follow these steps:

1: To enable to CPU monitor:

  •  Access the CPU monitor advanced URL at https://x.x.x.x:8082/Diagnostics/CPU_Monitor/statistics, and click start.  
    • NOTE: From here you can view it live, or stop the monitor.
  • For more information on how to setup the CPU monitor, see KB3605

NOTE: Once you have finished taking snapshots, be sure to turn the CPU monitor off.

2: To configure the ProxySG to take snapshots at 5 minute intervals:

  • In the Management Console > Maintenance tab > Service Information > Snapshots
  • Click "New"
  • Enter name: "CPU<proxyserialnumber> , or something meaningful, like CPU_Monitor_snap
  • Highlight the newly created snapshot and click "Edit" target: /Diagnostics/CPU_Monitor/Statistics/Advanced
  • Interval: 5 (as in 5 minutes)
  • Maximum number to store: 100
  • Check "Enabled" and click "OK"
  • Click the Apply button .
  • For more information on how to take snapshots , see KB3795

NOTE: To view the snapshot, wait five minutes, and then follow these steps:

  • In the Management Console > Maintenance tab > Service Information > Snapshots
  • Highlight the snapshot, created in the steps above, and press edit.
  • On the bottom left, click on View Snapshots.
  • Here you can download, or view the snapshot in the browser.
  • Each feature, enabled on your SG, will now be shown in this list if it is taking CPU cycles.

3: Taking a Full Core:

If it is not obvious that any one particular service is dominating the CPU, or the collected data is inconclusive, you might need to take a full memory coredump, as per this article KB3800.  Please consult with your Blue Coat Technical Support Engineer as to whether a full memory core is needed or not.

Once you have identified which feature is consuming you CPU, please review the list of articles below for a possible solution.  Depending on which which services are causing the high CPU, see these articles for more information on how to resolve them. 

High CPU in the  TCP/IP protocol stack      KB3606
HIgh CPU in the HTTP protocol stack     KB3640
High CPU is mostly in the Policy service   KB3641
High  CPU is mostly used by the Object Store service    KB3642
High  CPU usage is not taken up by any one service KB3638
High CPU usage in the ADN component.  TFA23
Smartfilter causing the high CPU utilization  TFA26
Unauthenticating ESS MIME clients causing high CPU.  KB3905
High CPU in TCP and DNS when using tcp-fast-finwait2-recycle KB4084
High CPU in policy evaluation caused by URL redirect  KB4186
High CPU on SG510 appliances, causing slow internet access. KB4397
High CPU in Policy evaluation- Websense issue. KB4462

NOTE: For a list of what Proxy SG versions contain fixes for high utilization issues, see FAQ903


Rate this Page

Please take a moment to complete this form to help us better serve you.

Did this document help answer your question?
 
 
If you are finished providing feedback, please click the RATE CONTENT button. Otherwise, please add more detail in the following text box and then click RATE CONTENT.
 
 

Your response will be used to improve our document content.

Ask a Question