Unable to get WU

Alles zum Projekt yoyo@home
Everything about the project yoyo@home
Nachricht
Autor
marsinph
PDA-Benutzer
PDA-Benutzer
Beiträge: 31
Registriert: 05.04.2018 09:08

Unable to get WU

#1 Ungelesener Beitrag von marsinph » 26.08.2018 07:44

Hello,
Since 25 august about 07:40 unable to get WU.
And very strange in log file :
26/08/2018 08:37:41 | yoyo@home | update requested by user
26/08/2018 08:37:44 | yoyo@home | Sending scheduler request: Requested by user.
26/08/2018 08:37:44 | yoyo@home | Requesting new tasks for CPU and NVIDIA GPU
26/08/2018 08:37:45 | yoyo@home | Scheduler request completed: got 0 new tasks
26/08/2018 08:37:45 | yoyo@home | No work sent
26/08/2018 08:37:45 | yoyo@home | No work is available for evolution@home
26/08/2018 08:37:45 | yoyo@home | No work is available for ecm
26/08/2018 08:37:45 | yoyo@home | No work is available for Perfect Cuboid
26/08/2018 08:37:45 | yoyo@home | No work is available for Siever
26/08/2018 08:37:45 | yoyo@home | (reached per-CPU limit of 2 tasks)

What is the latest line ?
It appers after server outage of yesterday PM
The same on all my hosts (all i7-2600k (8cores) and one i73820 (8cores).
All hosts run 8 WU at the same times (together) without any problems

Two days ago, I received 16 then 64 WU.
Now, no any !!!
What happens ??? Is it a new limitation from project ???
Considering SETIBZH and Formlula BOINC, if we are no able to get more than 2 task, why to run ?
Best regards from Belgium.

Benutzeravatar
yoyo
Vereinsvorstand
Vereinsvorstand
Beiträge: 8045
Registriert: 17.12.2002 14:09
Wohnort: Berlin
Kontaktdaten:

Re: Unable to get WU

#2 Ungelesener Beitrag von yoyo » 26.08.2018 08:58

Yes, I limited the number of tasks a host can have in progress to 2 per core to keep the server stable.

Which hostID is it? How many cores and how many tasks are on your host?
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

marsinph
PDA-Benutzer
PDA-Benutzer
Beiträge: 31
Registriert: 05.04.2018 09:08

Re: Unable to get WU

#3 Ungelesener Beitrag von marsinph » 26.08.2018 10:16

Hello, all my hosts are visible !
The 3 host below are all with CPU : I7-2600K, 3.8Ghz - 1 physical CPU / 8 cores / 8Gb RAM
All need a little less than 3 hours to run (all perfect cuboid).
Host 428425 8WU running (need 1hour to completion) and 8WU waiting to run
host426043 8WU running (will finish in about 30 minuts) 16WU waiting to run
host 427975 : 8WU running (will finish in 1 hour) and 6WU waiting !!!!


Then host (I7-3820K - 1 CPU / 8 cores) 8WU running ( 4 will finish in less than 1 hour - 4 will finish in 2 hours) - Waitings WU :62 (got one day ago)

I understand the limit (also to limit to much WU not returned in time,
but then it will say we can get a cache of maximum 3 hours (in my case).
It will say host need to be always connected !
What when outage or server problems ?
Then can you explain this host : 365004 , 1CPU/4 cores single thread who have about six hundred WU in cache since 05 august and no any crunched !!!
Or un E8400 dual core mono thread with 93 WU in cache since 15 august and no any crunched WU !!! !!!


I

Benutzeravatar
gemini8
Vereinsvorstand
Vereinsvorstand
Beiträge: 5898
Registriert: 31.05.2011 10:30
Wohnort: Hannover

Re: Unable to get WU

#4 Ungelesener Beitrag von gemini8 » 26.08.2018 11:01

Normally it isn't a problem to get more work.
yoyo just shortened the number of available work-units per core because of the formula boinc sprint to keep the server save.
You should have expected server problems rather before yoyo did this.

If you want to have more work for your machines I suggest you take on work from another project or fake more cores.
Gruß, Jens
- - - - - -
Lowend-User und Teilzeit-Cruncher

Bild Bild Bild
Bild

Benutzeravatar
yoyo
Vereinsvorstand
Vereinsvorstand
Beiträge: 8045
Registriert: 17.12.2002 14:09
Wohnort: Berlin
Kontaktdaten:

Re: Unable to get WU

#5 Ungelesener Beitrag von yoyo » 26.08.2018 16:36

As gemini explained this is a meassurement to keep the server save. I don't want that people buffer thousands of workunits and send them to the server shortly before the end of the race. therefore I limited the number of workunits in progress to 2 per core.
That you see no finished workunits on other hosts is, because I delte workunits very fast after they are validated.

With the 2 workunits per core, you always have some spare workunits if one finishes.
Where is the problem?
Flushing big buffers might lead to server outages, therefore the number is limited. You can always get new ones.
In such race it is most important to keep the server save.
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

Antworten

Zurück zu „Number crunching“