That's same with timeouts?
If so, you should report these results to Davis - they imply that 5.9.2a_beta is using a more generous timeout setting than the latest production version.
Yes, the same with timeouts. But interestingly enough, I am now seeing the large gaps (and "error finding aprs host") with 592a_beta and they are getting bigger and bigger. It looks like only about a dozen packets have made it to aprs in the last 24 (should be about 144) to cwop 1, 2 and 3. Another thread has started in the cwop forum by people recently noticing large gaps in their cwop data. I posted several tracert's to that thread in case it may be useful there.
It may not be a 5.9.3 problem afterall, but it is inconsistent enough to mess with my troubleshooting.