Results list is truncated at 16

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109409701194
RAC: 35114798
Topic 188122

If I log into my account and view my results I see exactly 16 even if I check from different machines, so it's not machine specific behaviour. I've got rather more than 16 entries now and all seems normal with the accumulation of credit. If new results are added, the oldest results drop off the bottom of the list and the visible list remains at 16. If I view other users results, I see them in pages of 20 as expected. I've looked around for some preference or setting that controls this but can't seem to find one.

Am I being totally dense and stupid? It's probably staring me right in the face but I don't see it. Perhaps it some sort of bug?? Someone please put me out of my misery, thanks!!

Cheers,
Gary.

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4267
Credit: 244933143
RAC: 16332

Results list is truncated at 16

I haven't looked into your particular case yet, but I think that you see just the normal behavior. You only see the Results that are still in the database, so somewhat "active". When the validator could determine a canonical Result and all the Results that have been sent out for a given Workunit have either returned or the deadline is over, this Workunit and all associated Results will evenatually be purged from the active database. With the current settings you should only come to see your Resuts of the last 1-2 weeks. If these happen to be 16 for your machines - well then.

BM

BM

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109409701194
RAC: 35114798

> will evenatually be purged

Message 6136 in response to message 6135

> will evenatually be purged from the active database. With the current settings
> you should only come to see your Resuts of the last 1-2 weeks. If these happen
> to be 16 for your machines - well then.
>
> BM
>

Thanks BM for the quick reply. Out of interest earlier today I'd been looking at older results and at the results of other computers that had processed the same WU as I had. I was sure I had looked through some lists of results that had gone back to about very early Feb, certainly before I joined on Feb 09. However at the moment all I see are results lists that start around Feb 13, just over two weeks ago. So my non-growing list is simply the result of older results being removed from the active database. No problem - thanks for the info.

The main reason I was looking back through old results was that when I joined on Feb 09, I set up a machine to process WUs 24/7, just before I departed on Feb 13 for an overseas trip. I checked on progress just before catching the flight and results were appearing regularly at just under 6 hour intervals. 32 hours later on Feb 14 there was a momentary power fluctuation which was so short that a number of computers actually survived whilst a number of them crashed. Unfortunately, the one running Boinc/EAH was one of those that crashed. There was no one to check on the boxes whilst I was away so the problem was only discovered when I returned. Power in this area is normally quite reliable. Machines stay up for many months on end usually.

When I rebooted the machine a new stdout.txt was created and the previous one was there as stdout.old. By looking through that file I can see regular results being uploaded on Feb 13 and 14 right up until the time of the crash. I had deliberately reduced the connect to network time from I think 1.0 days to 0.2 days on Feb 12 so as to reduce the cache I had to just a small number as it had previously been at about 8 WUs waiting to be processed. I reckon that the cache would have been depleted to about 1 or 2 WUs at the time of the crash.

As well as losing almost two weeks of processing time, I lost about 7 or so results that had been uploaded but had apparently not been through the second stage that I have seen others talk about. When I was able to restart, all of these results, although successful, were lost because they were now well out of time. I'm pretty sure that after reducing the connect to network time from 1.0 days to 0.2 days I forced an update to get the new value recognised. Why then had the completed and uploaded results not been through the next stage? Current ones are getting picked up fairly promptly now and I've even increased the time to 0.3 days to keep a bit more work on hand.

The main reason I'm documenting all of this is to perhaps see if there is anything that can be done to minimise the time between the result being uploaded (announced?) and the second stage (scheduling server?) coming along and grabbing it to insert it in the queue to be validated. From the number that I lost, this time difference must have been at least 2 days since I can see 8 results that were completed before the crash and were mostly ruled out of time when the machine was restarted on Feb 26. Interestingly one did sneak in as valid even though it was a full 7 days out of date.

Perhaps the scheduling server is more prompt these days in picking up results?

Regards,
Gary.

Cheers,
Gary.

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4267
Credit: 244933143
RAC: 16332

Roughly speaking the report

Roughly speaking the report of a result happens on the next server contact after the upload, so it should depend on your settings how often to contact the server.

There is a command line switch named -return_results_immediately that you can give at least to the command line client boinc_cli.exe (should work with boinc.exe, too, but I'm not sure) which will make it report Results immediatly after they are finished.

BM

BM

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5842
Credit: 109409701194
RAC: 35114798

> Roughly speaking the report

Message 6138 in response to message 6137

> Roughly speaking the report of a result happens on the next server contact
> after the upload, so it should depend on your settings how often to contact
> the server.
>
> There is a command line switch named -return_results_immediately that you can
> give at least to the command line client boinc_cli.exe (should work with
> boinc.exe, too, but I'm not sure) which will make it report Results immediatly
> after they are finished.
>
> BM

Thanks very much for your comments. I got that cli switch in another thread as well so I now have a solution. My contact period had been set at 1.0 days and had been reduced to 0.2 days approximately 2 days before the machine died, so there would have been plenty of server contact. Because my cache was running down I guess there wouldn't have been any new work being downloaded so that must have stopped the credit claim. Another proof of Murphy's Law :).

Cheers,
Gary.

Cheers,
Gary.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.