Common Problems

Did Not Become Root

The Traffic Manager needs to bind to privileged ports to balance network services like mail and HTTP (privileged ports are ports below number 1024). To do this you need to install, run the configure script and start the Traffic Manager as root.

In front-end fault tolerance mode, the Traffic Manager needs to raise and lower network interfaces. If you did not configure and start the software as root, the failover architecture does not function.

Note that the Traffic Manager runs as a non-privileged user, but it has to be configured as, and started as the root user.

For evaluation purposes, it is possible to configure and start the Traffic Manager without becoming root. However, in this case you will not be able to bind to privileged ports (see above), nor use fault tolerance.

Connection Refused

The most common configuration error for front ends is a bad DNS setup mapping the external name to the external IP addresses for your front-end machines. If you receive a “Connection Refused” message when connecting to your server through the front-end DNS name, it is likely that your DNS is not configured correctly.

To check your DNS configuration, you can use the host or nslookup commands:

$ host www.w3.org

www.w3.org has address 18.29.1.35

www.w3.org has address 18.7.14.127

www.w3.org has address 18.29.1.34

Verify that your Traffic Manager machines are listening on these IP addresses.

You can configure a virtual server to listen for traffic on these IP addresses explicitly. Then the Diagnose page will report an error if these IP addresses are not available to the Traffic Manager cluster.

Inappropriate Traffic IP Addresses Configured

If you configure a traffic IP group for front-end fault tolerance, the Traffic Manager will make a number of checks to ensure that the traffic IP addresses you select are sensible. However, you should also double-check that they are appropriate for your current network configuration. Common errors include selecting traffic IP addresses that are used elsewhere (including as permanent IP addresses on the Traffic Manager machines), IP addresses that do not lie in the subnets used by the Traffic Managers, or IP addresses that are not routable from elsewhere.

The Traffic Manager Drops Connection Before Protocol Begins

For server first protocols, when the Traffic Manager cannot contact any back ends for work for a particular port for server first protocols, it will have no choice other than to drop the connection with the client. If you connect to the correct port on the Traffic Manager machine (using telnet for instance), and get a successful connection (i.e. no “Connection Refused” message), but the connection drops almost immediately, the Traffic Manager is trying to talk to a back end but cannot find any.

Check the Diagnose > Event Log page for messages about the status of the back-end servers.

Web Server Returns Error 400

If on testing your traffic-managed Web site you only get errors of type Error 400 Bad Request, this may be because the Web site you are trying to access has not been configured to accept any other host headers (or aliases). In order to resolve this problem you can add a rule to set the right host header:

http.setHeader( "Host", "www.mysite.com" );

Wrong Port Number Configured

Another common problem is the wrong port number being balanced. If, for instance, your Web site is running on a port other than 80, you will need to balance that port number with the HTTP protocol rather than the default. If you are running an SSL-encrypted site, you should select the HTTPS protocol rather than HTTP, or SSL-decrypt your traffic. The default port for HTTPS is 443.

Note that it is possible to balance many distinct ports with one Traffic Manager installation, provided an appropriate protocol is selected in each case.

Running Out of File Descriptors

Your Traffic Manager uses operating system resources called “file descriptors” to manage each network connection to a client or server. There is a hard limit on the number of file descriptors the Traffic Manager can allocate (the limit applies per process, in other words, per CPU core), set in the operating system.

Under times of very high load, such as during a benchmark, you might see an error message indicating that you are “Running out of file descriptors”. In this case, you can increase the number of file descriptors that your Traffic Manager uses.

To increase the maximum number of file descriptors, log into the Admin UI and click System > Global Settings. In the “System Settings” section, increase the value of maxfds to a larger value. If you are already at the maximum that your operating system allows, refer to the following Knowledge Center article to ascertain whether this hard limit can be increased: https://kb.pulsesecure.net/articles/Pulse_Secure_Article/KB43935

Running Out of Disk Space

This section is applicable to appliance and cloud variants only.

Over time, your appliance can run low on disk space. For example, your system logs can become large if you have configured your Traffic Manager to produce detailed request log information. Additionally, archived software revisions (used by the Traffic Manager for roll back) might no longer be required.

The Traffic Manager warns you if disk space is running low through the Diagnose > Event Log and Diagnose > Cluster Diagnosis page. You can also view disk space usage at any time through the System > Traffic Managers page.

To free up disk space, click on "Free up some disk space" from the Wizards: drop-down menu in the top navigation bar. You can also run the wizard from the "Free Disk Space" link on the System > Traffic Managers page at any time, and from the Diagnose > Cluster Diagnosis page when a low disk space warning appears.

ATTENTION
This operation is irreversible. You should ensure you have created a backup of any files you need to keep before running the wizard. Note also that any "Technical Support Reports" you create afterwards contain only those logs generated since the wizard was run.