eValid™ -- Automated Web Quality Solution Browser-Based, Client-Side, Functional Testing & Validation, Load & Performance Tuning, Page Timing, Website Analysis, and Rich Internet Application Monitoring.

Introduction
This page gives basic recommendations about how to prevent cascading failure of eValid playbacks when eValid is used in monitoring mode, i.e. in a performance intensive, scheduled launch of large numbers of eValid playbacks.

• Cascading Failure Defined
  You may experience cascading failure due to a number of possibly different causes that occur either singly or in groups. In most cases cascading failure means that some of your launched Typical causes of cascading failure
  1. The resources available on your machine become insufficient to run all of the eValid playbacks you have scheduled.
  2. One or more eValid copies is frozen and can't be removed.
  3. Some other failure causes eValid copies to fail to launch, and these failures cause secondary "chain reaction" effects.

  This list is NOT exhaustive, it is important to emphasize. eValid, like any Windows process, can cause a wide variety of problems that can interfere with your monitoring activity.

• Foreground Interference
  If you use eValid while the machine that you're relying upon to monitor, and that use uses up too much resource, then any available safety factor may disappear, and you own use of eValid may precipitate a cascading failure mode. The solution to this kind of failure is to NOT USE evalid on a machine that is dedicated for playback only.

• -RT Switch
  The -RT switch will interrupt eValid operation at the specified wait interval, and it could be that this interruption causes unusual playback behavior. The -RT time limit should be set to be large enough so that the the limit includes nominal script playback time, plus an additional amount of "safety time". Otherwise the -RT switch itself could be the cause of cascading failure.

• Lock/Unlock Command
  Use of Lock and Unlock commands in monitoring mode is generally unwise. These two eValid commands are mainly intended for use in eValid server loading runs, to prevent two [or more] processes from interfering with each other when both are attempting to access the desktop.

  Even though the Lock command has a specified time limit (the default limit is 60 seconds) that may be too long and a subsequent attempt to run a test could be blocked by the Lock from a prior one being already in place.. Even if only for the +60 seconds, that may be too long and the failure of this one playback could cascade into failure of many/all subsequent playbacks. Best practice is to use Lock/Unlock as little as possible. Also, if you have multiple scripts that could interfere with each other, the Lock/Unlock operation needs to be in ALL of them.

• Operation on Virtual Environment
  We have heard of some problems when running eValid in monitoring mode in virtual environments. We don't know of any specific problem areas, but virtual environments are often just subtly enough different as to cause problems.

• Loader Failures
  Two messages that appear fairly often and seem to be related to problems when you are attempting to load an eValid instance are:
  • Failed to create empty document.
    This is probably due to the machine having insufficient RAM to accomodate the executable image.
  • An unsupported operation was attempted.
    This is probably due to the machine haveing insufficient disk space to write a file or into which to perform memory swapping operations.

  In both cases the solution will be to increase RAM, increase the available disk space, or reduce the number of eValid copies you are attempting to run.