LL assignments on slow PCs "die" shortly before finishing?

[Andi47]

I have seen this quite a few times in http://mersenne.org/primenet/status.txt

A LL assignment at the tailing end of the list, obviously on a slow PC (running since 400, or 500 or even 600+ days) which is regularely updated - until just one or two days before the exponent should be finished, and then.... nothing, the exponent gets overdue and finally expires.

What is going on? I can imagine maybe ONE unlucky GIMPSer who LLs an Exponent for nearly two years and then his 486 dies two days before finishing the test, but I can NOT imagine five or more people who lose their interest in GIMPS after more than one year of running a test just when their expo is 99,8% done.

Currently even If I do only a quick counting on the first two pages of status.txt I see more than five exponents at the tailing end which have run for more than 300 days, should have already finished, last update in Jan or Feb 2007 and thus overdue, at least one of them (14405561) last updated ONE day before it was due to finish....

Is it possible that some of these PCs run an older version of Prime95 which does not properly communicate a finished LL-test to the current version of Primenet?

[tha]

Originally Posted by Andi47

I have seen this quite a few times in http://mersenne.org/primenet/status.txt

I can testify that I have seen this phenomenon a number of times as well in the distant past. It is not something new. Hopefully these exponents do get checked in properly but do not get removed from the server or so. Just hoping that, don't know if it is the case.