I have 17 computers running the latest version of seti. After more than two weeks three of them do not show any credit allocated although the number of units completed between the three of them is over 200. I know that there are problems but is this usual?

Go to your seti @ home web "Your account" page.(http://setiweb.ssl.berkeley.edu/home.php)

[disabled for now] Click on the "results" link. This will show what WUs have been sent to your bevy of computers.

Click on one of the links under the "work unit ID" column.

This will show the various machines (including yours) that this particular WU was sent to for calculation.

Seti gives credit when 2 WUs are returned from separate machines, and have identical results.

You can find out which other machines (and thus other users) are holding onto your WUs and not returning results for them.
-------
Current problems:
A) A user joins seti with a bunch of machines, gets many WUs, becomes dissatisfied with S@H2, and then stops running S@H2. That users WUs are now in limbo.

B) A user using dialup and a computer connects and gathers MANY WUs (max 50/day) to process while not online. Only when user connects to internet again and remembers to tell boinc to send finished WUs do they get back to servers.

Until the WU's expiration date (drop dead date), the servers won't do anything with that other users copy of the WU.

When expiration is reached, servers put the WU back into the pool to be passed out to some other user's machine.
-------
Why:
With old S@H1, it is highly suspected that users would gather a number of WUs, calculate results, and then cross distribute their WUs to their many machines. Then each machine would submit all the results for every other machine as well as its own. 5 machines * 10 WU each = 50 WUs (normal). This cheater claims 250 WU credit.

With old S@H1 user could download S@H source code, modify it to get WUs, then not properly calculate the SCIENCE (in much shorter time), and then return WUs for credit.

No, that's not unheard of. Credit granting is dependent on several variables.

1- How quick are the other hosts the WU has been sent to?

2- How quick these other hosts will report the results. Will they do it manually, or let the scheduler pick them up automatically? If they go the automatic route, this can be accomplished two ways. It will either grab them the next time the host asks for more work, or around 24hours before the WUs deadline. If the host has a large cache, it could be a week or two before it asks for more work. If BOINC comes up against the deadline, this can also be two weeks away.

3- Has any host given the WU reset or detached, thus "flushing" the WU? If a WU has been "thrown out", the scheduler will wait until the deadline expires before adding it to the "pool" of WUs to be sent out again. This could take two weeks for the deadline, then more time depending on how many WUs are in the pool. Then we are back to numbers 1 and 2.

And finally,
4- How quick the validator processes the WU. Currently, there is a serious backlog the validator is chewing through.

So no, two weeks without credit granted on a few of your systems isn't cause for concern. You're still well within "spec", even assuming the validator was on top of the work load.