> Depending on how the FTP server fails, you might also see it in your > performance monitor...
Also true. OTOH, there are failure modes (such as the one Colin mentioned about getting unhappy with a minidisk) that won't show up in the console log or will show misleading symptoms (large buffer exhaustion, for example). > Running a Linux virtual machine with agents just to poll the FTP > server is only "no cost" when you have too much resources and spare > time. So run it on an external system, where the cost on Z is periodically another client session. Either way, you get a end-user perspective of the behavior you want to test rather than inferring it from secondary sources. There are also likely to be other services that need similar testing, which could benefit from a similar approach, and there is substantial existing code that can do the job without writing a bunch of stuff from scratch. It takes a lot less time to configure an existing tool than to gather the information to second-guess what is happening from the console.