bug tracking - Any good strategies for dealing with 'not reproducible' bugs? -

Many times you can get or report bugs for errors that are not 'reproducible' They may be reproducible on your computer or software project, but not on a vendor's system. Or takes the steps to reproduce the user, but you can not see the flaws locally. To make so many variations on this scenario, I think what I am trying to learn:

What is your company's policy regarding 'copyable' bugs? Shelve them, turn them off, ignore them? I sometimes see intermittent, non-duplicate bugs in the 3-rd party framework, and they are very close by the seller very quickly ... but they are real bugs.

Do you have any technique that helps you get these kinds of bug fixing? Generally I am receiving a system information report from the user, and steps to reproduce, search on keywords, and try to see patterns of any kind.

• Verify the steps used to generate errors

Often people or people reporting the error repeat the error, some will do wrong and will not end in the same state, even if they think they are. Try to walk with the reporting party. I had a user INSIST that the administrator privileges were not displaying correctly. I tried to redefine the error and unable to do it when we used to run it together, it was discovered that In that case, he was logging as a regular user.

• Verify the system / environment to generate error

I have received several 'immature' bugs and later It is only known that they are the enforcement of the X version of Mac OS (10.4) Safari. And it does not apply to browsers and rendering only, it can apply to anything; Whether other applications currently running, whether the user is RDP or local, admin or user, etc. ... make sure you call your environment as much as possible before making it inadequate.

• Collect screenshots and logs

Once you verify that the user is doing everything properly and There is still a bug, and that's what you are doing properly, and you are not getting a bug, so it's time to see what exactly you can do about it. Screenshots and logs are important You want to know exactly how this looks, and what was actually happening at that time.

It is possible that there may be some information in the log that you can reproduce on your system, and once you have an accurate scenario again, you may have an error outside the hiding To celebrate

Screenshots also help with it, because you can know that "x piece is correctly loaded, but this should not be done because it is dependent on y" and it gives you a signal

• Collect step-by-step details from the user

This is very common to blame users And they do not trust whatever they say (because they have a 'thing' to 'user' Es), but even if they are looking for, which will be able to describe not know, they still conduct their names that they are looking for. It may have included some minor mistakes that may occur a few minutes before the actual error, or perhaps it can usually slow down in some things, these clues to help you understand this. You can find which aspect is causing an error on your machine and not your own.

• Try alternative ways to generate errors

If everything fails, then the code Try looking at the section which is causing problems, and possibly use a reflector or a workaround if it is possible for you to create a situation where you are getting information before half (hopefully in the UAT) that Users to try out the approach Ask Ta, and see whether the error still occurs. Do you work best to alternate but similar methods that get the error in a different light so that you can better examine it.