I agree with David about understanding WHY something may solve a given problem, though of course there are often times when dealing with something new that it's easier said than done.

Personally I think if you can't understand it fully then the critical thing first of all is to understand the impact of the change. If I don't understand why it'll work, but it's something I can easily change back afterwards or which will have no negative effect on anything then I may just give it a punt and see what happens. If I find several possibly solutions, I'll work out which of them have the least impact (eg requiring a reboot, effecting other software, taking any services currently online offline etc), and then work through them from the least impacting onwards.

Finally, if I'm really hesitant about a suggested solution (I personally don't like making ANY changes I can't roll back if I can help it), then look into the history and reputation of the person offering the advice, do you get the impression that the person is very knowledgeable in the subject, or just a serial poster. Also, can you find any other people suggesting the same solution (while not being a copy and paste job) on other sites, with respondents confirming the fix worked for them?

At the end of the day it's really just another area of risk management like we deal with on a regular basis, especially those of us from a Sysadmin background. You can't always be 100% risk free, but the true test is to mitigate as many of those risks as you can before jumping onto the first solution you find.

Keith Langmead