Really getting pissed with Milkyway@home

[Markfw]

So a few days ago, I had no data. Then I added the 4th card. I was at a steady 7.5 mill a day with 3 cards (Titan V) and I added a 4th. Should be 10 mill a day. First I had no units, and 860 trying to upload. So then I added a spammer to ping the project from 2 different host. Got 6 mil. Then yesterday, nothing. Then today I had no units on one of the computers. Manually did an update, and got a few, but the daily stats just appeared. 1.8 mill instead of 10 ? I spent $3500 of these 4 cards just for milkyway@home, and now they are all but useless.

 

[Orange Kid]

DC in general seems to becoming over saturated with users nowadays.
Hang in there as Milkyway has had its share of problems lately.
What with all the advances in hardware, projects are getting more work done than they can handle I feel. Seems more and more projects are going by the wayside.

 

[Markfw]

OK, this is a little insane.... over 88 THOUSAND pending

 

[StefanR5R]

Sometimes there is a large validation backlog, for several weeks:
https://munin.kiska.pw/munin/Munin-Node/Munin-Node/workunits_milkyway.html

The validator will get to it at some point, in the long term.

PS, a hardware upgrade of the server is on the horizon:
message from Tom Donlon on September 21

 

[StefanR5R]

Oh wait,
the workunits of current valid results have "minimum quorum 2, initial replication 3". Newest of these workunits are from October 8, ~12:00 UTC apparently.

But the workunits of current pending results have "minimum quorum 1, initial replication 1". Oldest of these workunits seem to be from October 8, ~15:00 UTC, from a quick look through top hosts.

Does anybody know if this is normal for MilkyWay@home? Looks to me as if a mistake happened when the newer batch of workunits was generated. If so, this should be reported to the project admin.

 

[Kiska]

Sometimes there is a large validation backlog, for several weeks:
https://munin.kiska.pw/munin/Munin-Node/Munin-Node/workunits_milkyway.html

Prettier graph: https://grafana.kiska.pw/d/boinc/bo...om=1662989955477&to=1665581955477&viewPanel=3

 

[Kiska]

Oh wait,
the workunits of current valid results have "minimum quorum 2, initial replication 3". Newest of these workunits are from October 8, ~12:00 UTC apparently.

But the workunits of current pending results have "minimum quorum 1, initial replication 1". Oldest of these workunits seem to be from October 8, ~15:00 UTC, from a quick look through top hosts.

Does anybody know if this is normal for MilkyWay@home? Looks to me as if a mistake happened when the newer batch of workunits was generated. If so, this should be reported to the project admin.

I believe there may be a problem with adaptive replication and every host is now untrusted.

 

[Skillz]

Problems occur, no way around that anywhere. The admin(s) are aware of the issue(s) and they are doing what they can to get it fixed.

 

[VirtualLarry]

DC in general seems to becoming over saturated with users nowadays.

Ex-miners got GPUs, no-where to point them at. Some are discovering DC projects.