Current community support?
I've been tinkering with OpenSTA as a load test tool for our organization. Basically trying to give the open source tools a chance before defaulting to LoadRunner. The fact that it has an active forum here definitely gives it a few points in my book. However, it is troubling to see basic functionality problems(source line too long) that have not been addressed in over a year and a half. Couple that with references to IE 4/5/6/Netscape and this is a hard sell to management over something like Bad Boy/JMeter even if it has more functionality.
Is there any active community development going on for this tool? Seems that currently even the website is down.
Re: Current community support?
Yes, its somewhat active. I have a beta release that I plan to let go later this year. It has large variable support and built in url encode/decode. Essentially, its stuff I had added to work with viewstates. Sort of a .net/jsf version.
The IE6 references is really an indictment of the https handling. I routinely use the most current versions of all browsers with opensta, just not with https. Its a case of "there is a workaround" and the real fix requires a huge rewrite of the recording mechanism.
I'm the chief source of training and support these days for OpenSTA, so you might think I'm just "talking my book", but Opensta for all its warts is a very powerful tool. I use it on nearly all of my load test projects for customers and it works well. The demands OpenSTA places on servers (cpu and memory) is, and I'd wager a cold one, lower than any other tool out there. Lately I've moved it into the cloud and can generate 10s of thousands of users for next to nothing in cost.
For times when I can't use OpenSTA, I prefer using selenium RC from the cloud. I use the Browsermob web site to do so. Selenium and OpenSTA together make up a great tool kit. Seems what is very hard in one is easy to do with the other and vise versa. Between the two, I haven't had to walk away from any business due to lack of capability.
The scripts generated by OpenSTA take a fair amount of massaging to do things like correlating data from one request to the next (including viewstate handling), validating results from requests, and pacing. I wrote a script processor/editor that is option driven. It automates the common editing and reformatting tasks and also addresses silly annoying problems like "source line too long". If there is a repetitive editing task or a bug I got sick of working around, I coded a fix for it in my script editor.
Net/net, working with OpenSTA tends to be more labor intensive than other scripting languages but using a script processor removes most of the pain.
If you are serious about using it, I'd suggest a couple of days training and the productivity tool I have created. I mentor one or two companies a year in using OpenSTA and can discuss this further with you off line if there is interest
So, in the end, it really does work. Working with it is labor intensive, but there are tools that reduce this considerably. There are warts, and you need to know the work arounds. Its not sexy, but it is powerful and efficient.