My HT processor just finished, and uploaded, two experiments using hadsm3 4.02. Rather than running the two new 4.02 experiments that had already been downloaded I did a project reset so that hadsm3 4.03 would be downloaded and the new experiments would use that.

Attempt 1 failed when the line was dropped (no idea why) and as there seems to be no way to restart an aborted download I had to reset the project again.

Attempt 2 downloaded all the files but rejected some because of checksum errors.

I can\'t try again as my daily experiment quota has been exceeded!!

NB. I doubt _very_ much if the checksum errors ae caused by a problem with my line or modem, I\'ve downloaded a couple of large updates recently with no errors.

If you (Carl) keep the quota then it would seem to me that separating the hadsm download and the experiment download might be a good idea - you download hadsm and _then_ you download an experiment.

Changing the download code so that it tries, at least, to recover from a download error or allow a retry without a project reset would also be a good idea.

Ian

I've restarted the servers with "KeepAlive=On" so hopefully that will help. I've also reset "max results" for all hosts today in case you want to try again.

Thanks Carl, I've now got two experiments up and running under 4.03.

I did have one more aborted attempt caused, I think, because I was in the middle of a one hour deferrment. When I rebooted my machine BOINC connected to the server and downloaded a new experiment but made no attempt to download 4.03 itself.

In the end I waited until the deferrment had ended, deleted all files that had 4.03 in the filename, did a reset and everything then worked as expected.

Thanks again for the prompt response.

Ian

> In the end I waited until the deferrment had ended, deleted all files that had
> 4.03 in the filename, did a reset and everything then worked as expected.

cool, so I guess "KeepAlive = On" is the answer!