My VALID, MCM WU's that are still vislble:

My oldest VALID WU has a return time of "2023-10-24 01:13:34 UTC".

I still show 35326 items in the current listing down from the initial 36038 items from my posting on Nov 10, 2023 2:23:17 PM.

New WU's are still slow to me. So, I have not yet re-energized those systems previously powered off. My queue is typically set for zero additional days as i tend to get a lot (relatively speaking) of resends.

So, Speedy51, when you open your Results page and you go to the last page (in your case that would be 1700 divided by 25 - if you have 25 items per page - which equals 68) of all your selected Valid results, you would see nothing out of the ordinary

My oldest Valid result (currently from page 669) is:

Result name         Status Sent time           Due / Return time   CPUtime/Elapsed Claimed/Granted
MCM1_0206648_5472_0 Valid  2023-10-23 23:42:31 2023-10-24 14:09:51    2.52/2.54       75.4/75

Adri

Further to Adri and Mike's observations:

My oldest is MCM1_0206667_9817_0 (WU 407998880) -- it was returned on 2023-10-24 at 02:42:57 (UTC)

If the data in the ModTime field returned by the old API can be believed, it was validated at 02:43:16 the same day.

I see no sign that assimilation is happening at present, so there's an associated worry. According to the Global Statistics History MCM1 is seeing over a million returned (and credited) results a day at present, so that's over 500,000 workunits a day. With 20 or more days where almost no WUs have been assimilated and purged, that's an enormous backlog to clear! A standard assimilator does up to 1000 WUs at a time (but how long that might take is an unknown)... Given how far it got clearing out stuff a few days ago, I suspect it might take a week or more of uninterrupted running! I leave any further calculations regarding this as an exercise for the reader :-)

Cheers - Al

Unixchick ,
Yup -- that's one of the possibilities I considered as "requires drastic action" in my post in the Chat Room thread on 14th November. And it is having an effect -- collecting or refereshing statistics that come direct from the BOINC database can be quite sluggish (whether via API or web pages...)
The "Statistics History" for MCM1 reports nearly 30 million results returned since 24th October (inclusive of that day); that suggests somewhere in the vicinity of 15 million WUs stuck waiting for assimilation! That's a lot of items to scan looking for WUs ready for assimilation[*1].

Unfortunately, the solution is likely to involve either shutting off the work-flow to and from users (to avoid further build-up) or effectively taking the whole service off-line for long enough to do something (such as multiple MCM1 assimilators if not already in use?) to clear some of the backlog[*2] with (hopefully) reduced stress on the database. For all we know, the techs may already be juggling various "scheduler" processes in an effort to sort things out, but there doesn't seem to be much progress :-(

Cheers - Al.

*1 The much smaller number of OPNG tasks live on the database at any time potentially makes for a much quicker scan -- it appears that OPNG tasks still clear within a couple of days of validation, so the database can still cope at present if circumstances are favourable!

*2 However, imagine the outcry from a fair proportion of the user base if anything interrupts work flow, even if it's an essential action! (Totally unlike the ARP1 situation, where it's obvious that the blame should not be laid on WCG...)

[Edited regarding sluggish statistical data delivery, and on "blame"]

TigerLily,

Thanks for the confirmation of the issue and the efforts to resolve it.

I have visions of many megabytes of assimilator log files and the accompanying headaches...

Cheers - Al.