No work is available for ecm

Fehler und Wünsche zum Projekt yoyo@home
Bugs and wishes for the project yoyo@home
Nachricht
Autor
Dunckx
PDA-Benutzer
PDA-Benutzer
Beiträge: 45
Registriert: 12.11.2014 09:26

No work is available for ecm

#1 Ungelesener Beitrag von Dunckx » 25.09.2017 11:23

I noticed yesterday evening that the CPU usage on my system was less than 100%. Upon investigation, the task list was indeed out of fresh WU, so I temporarily altered the network access time restrictions and tried downloading more. This has worked after a fashion, but today again my pc is short of work. I tried altering the preferences setting to download two days worth of work, but it has only eight WU ready to start and this will only last two and a half hours.

I decided to look at the logfile settings and checked the boxes for task debug and work_fetch debug. The last part of the logfile is:

25/09/2017 10:54:44 | yoyo@home | [work_fetch] REC 7117.675 prio -1.007 can request work
25/09/2017 10:54:44 | | [work_fetch] --- state for CPU ---
25/09/2017 10:54:44 | | [work_fetch] shortfall 1597763.48 nidle 0.00 saturated 11656.96 busy 0.00
25/09/2017 10:54:44 | yoyo@home | [work_fetch] share 1.000
25/09/2017 10:54:44 | | [work_fetch] --- state for NVIDIA GPU ---
25/09/2017 10:54:44 | | [work_fetch] shortfall 216000.00 nidle 1.00 saturated 0.00 busy 0.00
25/09/2017 10:54:44 | yoyo@home | [work_fetch] share 0.000 project is backed off (resource backoff: 778.43, inc 600.00)
25/09/2017 10:54:44 | | [work_fetch] ------- end work fetch state -------
25/09/2017 10:54:44 | yoyo@home | [work_fetch] set_request() for CPU: ninst 8 nused_total 16.00 nidle_now 0.00 fetch share 1.00 req_inst 0.00 req_secs 1597763.48
25/09/2017 10:54:44 | yoyo@home | [work_fetch] set_request() for NVIDIA GPU: ninst 1 nused_total 0.00 nidle_now 1.00 fetch share 0.00 req_inst 1.00 req_secs 216000.00
25/09/2017 10:54:44 | yoyo@home | [work_fetch] request: CPU (1597763.48 sec, 0.00 inst) NVIDIA GPU (216000.00 sec, 1.00 inst)
25/09/2017 10:54:44 | yoyo@home | Sending scheduler request: To fetch work.
25/09/2017 10:54:44 | yoyo@home | Requesting new tasks for CPU and NVIDIA GPU
25/09/2017 10:54:47 | yoyo@home | Scheduler request completed: got 0 new tasks
25/09/2017 10:54:47 | yoyo@home | No work sent
25/09/2017 10:54:47 | yoyo@home | No work is available for ecm
25/09/2017 10:54:47 | yoyo@home | (reached per-CPU limit of 2 tasks)
25/09/2017 10:54:47 | yoyo@home | [work_fetch] backing off CPU 513 sec
25/09/2017 10:54:47 | | [work_fetch] Request work fetch: RPC complete
25/09/2017 10:54:52 | | [work_fetch] ------- start work fetch state -------
25/09/2017 10:54:52 | | [work_fetch] target work buffer: 129600.00 + 86400.00 sec
25/09/2017 10:54:52 | | [work_fetch] --- project states ---
25/09/2017 10:54:52 | yoyo@home | [work_fetch] REC 7117.675 prio -0.007 can't request work: scheduler RPC backoff (2.63 sec)
25/09/2017 10:54:52 | | [work_fetch] --- state for CPU ---
25/09/2017 10:54:52 | | [work_fetch] shortfall 1597778.37 nidle 0.00 saturated 11656.96 busy 0.00
25/09/2017 10:54:52 | yoyo@home | [work_fetch] share 0.000 project is backed off (resource backoff: 507.87, inc 600.00)
25/09/2017 10:54:52 | | [work_fetch] --- state for NVIDIA GPU ---
25/09/2017 10:54:52 | | [work_fetch] shortfall 216000.00 nidle 1.00 saturated 0.00 busy 0.00
25/09/2017 10:54:52 | yoyo@home | [work_fetch] share 0.000 project is backed off (resource backoff: 770.86, inc 600.00)
25/09/2017 10:54:52 | | [work_fetch] ------- end work fetch state -------
25/09/2017 10:54:52 | | [work_fetch] No project chosen for work fetch
25/09/2017 10:54:55 | | [work_fetch] Request work fetch: Backoff ended for yoyo@home
25/09/2017 10:54:57 | | [work_fetch] ------- start work fetch state -------
25/09/2017 10:54:57 | | [work_fetch] target work buffer: 129600.00 + 86400.00 sec
25/09/2017 10:54:57 | | [work_fetch] --- project states ---
25/09/2017 10:54:57 | yoyo@home | [work_fetch] REC 7117.675 prio -1.007 can request work
25/09/2017 10:54:57 | | [work_fetch] --- state for CPU ---
25/09/2017 10:54:57 | | [work_fetch] shortfall 1597788.43 nidle 0.00 saturated 11656.96 busy 0.00
25/09/2017 10:54:57 | yoyo@home | [work_fetch] share 0.000 project is backed off (resource backoff: 502.75, inc 600.00)
25/09/2017 10:54:57 | | [work_fetch] --- state for NVIDIA GPU ---
25/09/2017 10:54:57 | | [work_fetch] shortfall 216000.00 nidle 1.00 saturated 0.00 busy 0.00
25/09/2017 10:54:57 | yoyo@home | [work_fetch] share 0.000 project is backed off (resource backoff: 765.74, inc 600.00)
25/09/2017 10:54:57 | | [work_fetch] ------- end work fetch state -------
25/09/2017 10:54:57 | | [work_fetch] No project chosen for work fetch

Aside from lines referring to the project being "backed off" and a rather worrying line saying "No project chosen for work fetch", there is " No work is available for ecm".

I have tried altering my BOINC settings to ask for two days worth of WU, I have tried rebooting my pc, restarting BOINC and getting it to reread the config and preferences file, but no joy. My tasks list still has only eight WU to go, which it will finish around two hours from now. It doesn't seem to be anything to do with my pc or the BOINC preferences. Is there a server problem?

Dunckx

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

Re: No work is available for ecm

#2 Ungelesener Beitrag von yoyo » 25.09.2017 11:26

25/09/2017 10:54:47 | yoyo@home | (reached per-CPU limit of 2 tasks)

You have already 2 tasks per CPU on your system.
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

Dunckx
PDA-Benutzer
PDA-Benutzer
Beiträge: 45
Registriert: 12.11.2014 09:26

Re: No work is available for ecm

#3 Ungelesener Beitrag von Dunckx » 25.09.2017 12:12

The problem is that the task list usually has between ten and a hundred WU in it depending on the runtime estimate of the WU. At the moment, it has only enough WU for the next couple of hours (8 WU ready to start and only 2.5 hours long) and it should have enough for at least a day and a half. For 2.5 hour WU that should be around ten per cpu, or 80 in total. As you can see from the screenshot, it is set to download enough work for the next 1.5 days, and I have just set it to download enough for yet another day - "Store up to an additional 1 days of work".

My system is supposed to go online and report to the Yoyo server between 0200 and 0659 each morning (this is set under "Daily schedules" and has been doing this faultlessly for a long time. All of a sudden, this has broken, and I do not understand why. In those five hours, it should report a few times, download enough WU for the next day and a half, then go offline until the next morning when it should report on what it has crunched for the previous day.

It isn't doing that anymore, and when this first appeared last evening I wondered what was going on. I just chanced to look at the Windows Task Manager, saw that CPU usage was less than 100%, and found only a couple of WU ready to start. I changed the daily schedule settings to allow the pc to download more work, saw that it only got enough for a few hours, thought that it would be enough to last until 0200 this morning when it would report back to the server and get more WU for the next day and a half as it usually does (did), and I then changed the daily schedule settings back to a network access time of 0200-0659. It has indeed downloaded some work from this morning's network time of 0200-0659, but only enough WU for a few hours, not enough for a day and a half as it usually does (did).

Today I have changed the "Daily schedules" settings a couple of times in order to get more WU, but I shouldn't need to do that. There ought to be enough downloaded for 36 hours or so from a single request to the server, which should happen between 0200 and 0659 as it has for a long time.

Very puzzled!
Dunckx
Dateianhänge
WU 250917 1140.jpg
WU 250917 1140.jpg (168.52 KiB) 4625 mal betrachtet

Dunckx
PDA-Benutzer
PDA-Benutzer
Beiträge: 45
Registriert: 12.11.2014 09:26

Re: No work is available for ecm

#4 Ungelesener Beitrag von Dunckx » 25.09.2017 16:16

OK, I've let the pc finish crunching the earlier WU, then reset the project, and again it is only downloading sufficient for a couple of hours. The cpu is currently busy with eight and has only eight more in the pipeline instead of enough for a whole day plus, and these WU are only an hour each. It will be done crunching in two hours instead of 24 plus.

The logfile (see below) has an entry stating "25/09/2017 15:58:42 | | [work_fetch] target work buffer: 86400.00 + 86400.00 sec" and each period of 86400 seconds is one day, i.e. it is trying to get two days worth of WU, which is what it should be getting. It isn't getting it (me neither!)

Dunckx
Dateianhänge
WU 250917 1600.jpg
WU 250917 1600.jpg (205.74 KiB) 4610 mal betrachtet

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

Re: No work is available for ecm

#5 Ungelesener Beitrag von yoyo » 25.09.2017 16:38

The reason for the change is on my side. I changed the server settings to allow only 2 Tasks in progress per CPU, because I have some issues with database load and want to reduce the amount of workunits in the DB.
Can't you change the settings to allow network traffic the whole day?

We can also direct discuss it in Slack: https://rechenkraftnet.signup.team/
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

Dunckx
PDA-Benutzer
PDA-Benutzer
Beiträge: 45
Registriert: 12.11.2014 09:26

Re: No work is available for ecm

#6 Ungelesener Beitrag von Dunckx » 25.09.2017 19:26

Ah, OK, if it's an "official" change I can live with that. I thought something at my end had gone bang and I was frantically trying to find out what, without success. Now I know why and can heave a sigh of relief. It isn't malware and I haven't been rootkitted. You did have me worried though. Please give a warning announcement if there's something of that nature again. Thanks!

It is a bit of a pain when something to do with computers suddenly changes its behaviour without an apparent cause. Yes, I know. Windows is like that...;-)

Dunckx

Antworten

Zurück zu „Fehler, Wünsche / Bugs, Wishes“