Agreed, but there is another possible reason validation can be delayed or not necessarily correlate well with return times, and that is transitioner backlog. Fortunately, when the system is running properly that should be relatively uncommon :-)

That comment has given me another project to run against my fairly substantial backlog of result data, as it is not my understanding of how things work, and I'm as interested in the method as in the outcome :-) -- opinion being just that, I will try to get some confirmation one way way or the other.

I've already ploughed through the source code for what I believe is the current BOINC server version on GitHub and can find absolutely nothing to support the "stored validation time" hypothesis :-) -- that said, although I seem to recall frequent "we can't do X because BOINC doesn't support it" comments from WCG/IBM in the past, it doesn't guarantee that no changes to core functionality have been made by WCG at some point[*1] :-)

It'll take me a day or so to code up a script to do the analysis, but I'll try to compare the first "Valid" ModTime seen for each result (which is probably the validation time if the delete state is 0) against the delete state 2 time (if I have it) and the sample time of the first data collection that didn't show the WU (which is the nearest I can get to a purge time)

I already know there will be some oddities in there because of the [original] discussion about how many assimilators were running -- as a result of the "just one out of four" period, assimilation, file deletion and purging is ahead for WUs with IDs for which the modulo 4 value is 1 :-) I already have results purged for as recently as mid-February 2024, and they all have WU IDs that match that constraint! Because of this, I'm going to try to dig out some data from before the assimilator problems and have a look at that first!

I'm still behind you here :-) -- I still have a few waiting from 23 December and about 100 from 6th and 7th January, and all of them have WU IDs that give remainder 0 when divided by 4; my overall statistics also suggest that that assimilator might have been a bit behind...

Cheers - Al.

*1 As a recent example, It is unclear whether the assimilator patches to avoid lock-ups/crashes were only made to the WCG-specific assimilation support routines or required changes to the BOINC assimilator wrapper as well. I'd hope the former, but that left me wondering why the one assimilator that kept going for quite a while managed to keep working upwards through WU IDs whilst leaving WUs it could have handled; perhaps it had already been patched (or WCG had a "start at this WU ID" mod (or similar) in place already...)

Getting there. For the first time in about a month, my results loaded. Looks like a few from October are left, but only a few. Then, from 12/24/2023 my results are still being purged.

Only 253,490 left!...... I see progress. I think the system is working well, it is just having to catch up.

Yes, even I, now only running a relatively slow Laptop, have trouble to get the 50 or so MCM1 tasks, that it needs/24 hours. Let's hope the team fix that issue, before they leave for the weekend.

The MCM WUs are few and far between. Don't think I saw a single one today.

My results list is shorter though. My oldest one is now Jan 12.

rough guess... 6 days results processed in a day. Sure some of you will have a better estimate. maybe another 10 days of this minimum ?? maybe some extra time to clean up some of the older stragglers some of you still have. good to see progress, but I do miss doing WUs.

Mine is purged to 20 Jan ! Still making slow progress.

I managed to get a new MCM task. I feel like I won the lottery. I didn't get any yesterday.