This should be part of your test plan, in particular, your 'Entry Criteria'. This is what must occur before you will start your testing, and includes everything which must be so in order for your testing to provide cost effective benefit and to fit within any limitations (such as schedule) you must operate under.
You must have the hardware, software and personnel needed to do the test available. If not, you cannot test.
You must have your test plan (including description of the testing you will be doing) reviewed and approved. If not, then you will be testing 'blind' and possibly not doing things which are valid or complete.
All previous development and test activity must be completed (best) or at the specified level of completion. If not, the odds are that either the thing you are testing will 'not work', or the tests you do complete will have to be redone when the 'real' thing is finally ready for test.
There must be no open problems which would significantly impact your testing. Obviously, a typo in a message should not hold up testing, but if the main function 'mostly does not work' or 'damages the test system', then again you risk being unable to test, or having to retest.
Anything specific to your circumstances to ensure that when the test starts, it has a valid subject and the testing to be done is valid. (In particular, anything which bit you in previous tests)
John