Panic Mode On (110) Server Problems?

Message boards : Number crunching : Panic Mode On (110) Server Problems?
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 37 · Next

AuthorMessage
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1916238 - Posted: 1 Feb 2018, 5:47:51 UTC - in response to Message 1916183.  

Are you sure your computer "Location" (Needs translation, LOL) matches that for which the CPU tasks are disabled?
Other locations have it turned on.
hi, i believe that i'm undestanding these settings like :

in Default or Work parameters:
for only AP app, use CPU and GPU 10 days cache ( 1 CPU AP / day ( +24h to crunch it ) , a lot of more for GPU ) no work for other app ( MB )
in Home parameters:
for MB or AP app, Use only GPU 1day cache

am i right ?
This page shows your Computer Location at the bottom of the screen. Only you can see what is selected, we can not. It must match the Location set for No CPU Use. Or modify the location that DOES match.
ID: 1916238 · Report as offensive
Profile Kissagogo27 Special Project $75 donor
Avatar

Send message
Joined: 6 Nov 99
Posts: 716
Credit: 8,032,827
RAC: 62
France
Message 1916282 - Posted: 1 Feb 2018, 13:52:22 UTC

Hi Brent , Grant , thanks a lot for your answers .

with this picture , you can see wich location is set for my computer http://nsa39.casimages.com/img/2018/01/31/180131073958773848.jpg

look at the top "Localisation par défaut de l'ordinateur: ---" " --- " means default location actualy set

i could stay in Home location now because of lacking in AP wu , but normaly i do switching between Home and Default or Work for catching few AP wu while my arecibo MB cache is emptying ..

the situation is not to crunch or not CPU WU , it's to not downloading MB CPU WU with uncheck CPU use for MB Wu ( aka home location settings ) > not downloading CPU wu but it does and
with 1 day cache , it must download only less than a decade CPU wu but do a lot more .

i presume i understand all of these parameters like i want to made them functioning like i want:

Home location > download only 1 day cache of only GPU WU for MB app or AP app
Default(---) or Work location > download only 10days cache for CPU and GPU WU for only AP app ..

for tests, i can stay in home location and see if the situation for CPU WU will change ...
ID: 1916282 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22535
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1916285 - Posted: 1 Feb 2018, 14:06:09 UTC

You cannot have different cache sizes for the different locations, you have manually set them for each computer - not via the website.
Physically moving a computer from one place to another will not change it's logical location.

Your "default" location (the first column in your picture) will download work for both your CPU & GPU, as will your "Work" {travail} (last column)
Only your "Home" {maison} location (second column) will not download CPU tasks.
You have no options set for "School" {ecole}

Tasks that have been downloaded will be run, so this trial could take some time to ensure that you are only getting the work you think you want. Given the rarity of AP work, I would suggest that you may do better to just set "accept AP" work for all venues, and take any that come your way as a bonus!
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1916285 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1916289 - Posted: 1 Feb 2018, 14:44:13 UTC - in response to Message 1916246.  


. . You must be young ... I am too old to wait a year :)
Stephen
???

Well, if you're older than 62, then I'm young, Otherwise not :-)


. . OK then ... you are young! :)

Stephen

:)
ID: 1916289 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51478
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1916291 - Posted: 1 Feb 2018, 14:52:08 UTC - in response to Message 1916289.  


. . You must be young ... I am too old to wait a year :)
Stephen
???

Well, if you're older than 62, then I'm young, Otherwise not :-)


. . OK then ... you are young! :)

Stephen

:)

Lot of us old timers here on Seti......LOL.
"Time is simply the mechanism that keeps everything from happening all at once."

ID: 1916291 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1916294 - Posted: 1 Feb 2018, 15:22:57 UTC - in response to Message 1916282.  

OK, I think I have this straight now :D

You want:
- 10 day cache, CPU + GPU, AP Only
- 1 day cache, GPU Only, AP + MB

The problem is with how (and when) the server tells the client about the change. When you tell the server you want to change to GPU only, the client does know that - so the next request will be asking for CPU + GPU tasks. I have run into this before, and it's a pain!

The way around it is to make the server change DURING the 5 minute timer, and then do a manual update BEFORE the timer expires. Since it is less than 5 minutes between server contacts it will NOT send you work for this request, but it will update the client settings from the server.
You could do - manual update, change server setting, manual update. That should work for you.

I'm not sure about honoring the cache size though, I have never tried that part.
But you would have to make sure the CLIENT is set to use WEB Preferences.
ID: 1916294 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1916309 - Posted: 1 Feb 2018, 16:41:27 UTC - in response to Message 1916294.  

Brent is correct. To get the servers to look at a new or different configuration, you have to update mid-normal connection interval.

I've never been able to get the cache size to work except globally. So 90% of Einstein work received at each request is dumped immediately. The venue mechanism does not work well unless you have a different venue on each computer and only do a single project on that computer.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1916309 · Report as offensive
Profile mr.mac52
Avatar

Send message
Joined: 18 Mar 03
Posts: 67
Credit: 245,882,461
RAC: 0
United States
Message 1916310 - Posted: 1 Feb 2018, 16:48:20 UTC - in response to Message 1916246.  

My definition of Old is my age plus 10 years...
ID: 1916310 · Report as offensive
Profile JaundicedEye
Avatar

Send message
Joined: 14 Mar 12
Posts: 5375
Credit: 30,870,693
RAC: 1
United States
Message 1916320 - Posted: 1 Feb 2018, 18:08:09 UTC - in response to Message 1916318.  

My definition of Old is my age plus 10 years...

Mine was 20 years older.

At 10 - 30 is old.
At 20 - 40 is old.
At 30 - 50 is old.
At...

But now, at my age. 10 years more is old. Therefore, 80 is old.

+1

"Sour Grapes make a bitter Whine." <(0)>
ID: 1916320 · Report as offensive
Profile Jimbocous Project Donor
Volunteer tester
Avatar

Send message
Joined: 1 Apr 13
Posts: 1856
Credit: 268,616,081
RAC: 1,349
United States
Message 1916325 - Posted: 1 Feb 2018, 18:31:21 UTC - in response to Message 1916309.  

... So 90% of Einstein work received at each request is dumped immediately ...

I solved that by setting Einstein to allow new work, but with a Resource Share of 0. Daily caches aren't considered, and I use Venues only on SETI.
That way, when I run out of S@H tasks, it gets exactly 1 Einstein job, runs it, uploads it and gets one more. Keeps the GPUs working, but with no excess or abandons to deal with. Ymmv.
ID: 1916325 · Report as offensive
Profile Ghan-buri-Ghan Mike

Send message
Joined: 27 Dec 15
Posts: 123
Credit: 92,602,985
RAC: 172
United States
Message 1916332 - Posted: 1 Feb 2018, 19:23:58 UTC

My ages:

Chronologically: 58
Emotionally: 6
Age of liver: 103+/-
ID: 1916332 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1916333 - Posted: 1 Feb 2018, 19:24:41 UTC - in response to Message 1916325.  

But I have never run Einstein as a emergency backup project. I have always run it and MilkyWay as secondary projects to Seti with 5-20% resource share depending on mood or work availability.

So I have just come to accept that Einstein will overdeliver and I just abort tasks as needed for the work that won't get crunched by deadline.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1916333 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1916337 - Posted: 1 Feb 2018, 19:48:19 UTC - in response to Message 1916333.  

But I have never run Einstein as a emergency backup project. I have always run it and MilkyWay as secondary projects to Seti with 5-20% resource share depending on mood or work availability.

So I have just come to accept that Einstein will overdeliver and I just abort tasks as needed for the work that won't get crunched by deadline.
Have you ever used the Event Log to view exactly what is being requested and allocated? This is pretty typical for me:

01/02/2018 19:30:18 | Einstein@Home | [sched_op] Starting scheduler request
01/02/2018 19:30:18 | Einstein@Home | Sending scheduler request: To fetch work.
01/02/2018 19:30:18 | Einstein@Home | Reporting 2 completed tasks
01/02/2018 19:30:18 | Einstein@Home | Requesting new tasks for Intel GPU
01/02/2018 19:30:18 | Einstein@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
01/02/2018 19:30:18 | Einstein@Home | [sched_op] NVIDIA GPU work request: 0.00 seconds; 0.00 devices
01/02/2018 19:30:18 | Einstein@Home | [sched_op] Intel GPU work request: 875.28 seconds; 0.00 devices
01/02/2018 19:30:20 | Einstein@Home | Scheduler request completed: got 2 new tasks
01/02/2018 19:30:20 | Einstein@Home | [sched_op] Server version 611
01/02/2018 19:30:20 | Einstein@Home | Project requested delay of 60 seconds
01/02/2018 19:30:20 | Einstein@Home | [sched_op] estimated total CPU task duration: 0 seconds
01/02/2018 19:30:20 | Einstein@Home | [sched_op] estimated total NVIDIA GPU task duration: 0 seconds
01/02/2018 19:30:20 | Einstein@Home | [sched_op] estimated total Intel GPU task duration: 1260 seconds
Obviously, my choices are likely to be different to yours, but my experience with Einstein, and indeed with all BOINC projects, is that "you get what you ask for", and not substantially greater. That's how it's supposed to work.
ID: 1916337 · Report as offensive
Profile Jimbocous Project Donor
Volunteer tester
Avatar

Send message
Joined: 1 Apr 13
Posts: 1856
Credit: 268,616,081
RAC: 1,349
United States
Message 1916346 - Posted: 1 Feb 2018, 20:25:50 UTC - in response to Message 1916337.  
Last modified: 1 Feb 2018, 20:26:44 UTC

Obviously, my choices are likely to be different to yours, but my experience with Einstein, and indeed with all BOINC projects, is that "you get what you ask for", and not substantially greater. That's how it's supposed to work.

Thought I heard somewhere that it will over-deliver until the "debt" is paid and actuals are matching what percentage was selected? If so, and that percentage is never actually reached won't it always over-deliver? Afaik, that's a BOINC thing, not unique to each project? Or am I totally off the mark here?
ID: 1916346 · Report as offensive
Profile Jimbocous Project Donor
Volunteer tester
Avatar

Send message
Joined: 1 Apr 13
Posts: 1856
Credit: 268,616,081
RAC: 1,349
United States
Message 1916347 - Posted: 1 Feb 2018, 20:28:47 UTC - in response to Message 1916332.  

My ages:

Chronologically: 58
Emotionally: 6
Age of liver: 103+/-

Sadly, I seem to have you by 4 in each area except the center ... (though I hope to increase there as I enter month 3 of sobriety)
ID: 1916347 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1916351 - Posted: 1 Feb 2018, 20:38:32 UTC - in response to Message 1916346.  

Obviously, my choices are likely to be different to yours, but my experience with Einstein, and indeed with all BOINC projects, is that "you get what you ask for", and not substantially greater. That's how it's supposed to work.
Thought I heard somewhere that it will over-deliver until the "debt" is paid and actuals are matching what percentage was selected? If so, and that percentage is never actually reached won't it always over-deliver? Afaik, that's a BOINC thing, not unique to each project? Or am I totally off the mark here?
That might be the case if you are thinking about 'over-deliver' as a long term result of allowing both projects to "request at will" for an extended period of time. If you've been exclusively requesting from one of the two projects for a while, BOINC will preferentially request from the other one when both are allowed, until resource shares are better balanced. [The actual word 'debt' hasn't been used in BOINC code since 2010, but the principle remains in place by other means]

My point was that Einstein won't over-supply as a result of a single work request: it'll only top up the cache by whatever the current shortfall is. Of course, if you ask for 10 days, and SETI will only ever supply you with 8 hours because of task limits, you might be surprised to get what you wish for ;-)
ID: 1916351 · Report as offensive
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1916353 - Posted: 1 Feb 2018, 20:55:08 UTC - in response to Message 1916351.  

Richard, you are correct. I am getting more work than I want because of the global hard disk allocation and also the global work days request. Einstein is sending me exactly what I ask for . . . . 2 days worth. But they overestimate what I can process in 2 days because the resource share is 5% of what I ask for Seti.

I have tried reducing my global work request down to 0.5 days and that does in fact reduce the amount of work that Einstein sends me. But it also for some reason limits the work I get from Seti which then can't keep my caches filled. It is not supposed to work the way it does. I just accept the fact at this time and figure out what in fact does work.

So I choose the lesser of two evils. I dump the excess of work that that I won't process in time from Einstein after each work request and set NNT until I have processed what I have on board and then repeat the cycle. This way I can keep my Seti caches full and still process work for Einstein. MilkyWay has never been a problem for me because of how they allocate a fixed amount of work at any time. That is a set and forget project. I wish Einstein would follow suit with a hard limit of the number of tasks on board like MilkyWay.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1916353 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1916365 - Posted: 1 Feb 2018, 22:07:37 UTC - in response to Message 1916310.  

My definition of Old is my age plus 10 years...


. . Which would be? :)

Stephen

??
ID: 1916365 · Report as offensive
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1916366 - Posted: 1 Feb 2018, 22:09:59 UTC - in response to Message 1916332.  

My ages:

Chronologically: 58
Emotionally: 6
Age of liver: 103+/-


. . Oh I know that one ....

:(
ID: 1916366 · Report as offensive
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1916369 - Posted: 1 Feb 2018, 22:21:26 UTC - in response to Message 1916353.  
Last modified: 1 Feb 2018, 22:23:02 UTC

I have tried reducing my global work request down to 0.5 days and that does in fact reduce the amount of work that Einstein sends me. But it also for some reason limits the work I get from Seti which then can't keep my caches filled. It is not supposed to work the way it does. I just accept the fact at this time and figure out what in fact does work.
But that is also 0.5 days of 80% if you are trying to do 20% elsewhere ...
ID: 1916369 · Report as offensive
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 37 · Next

Message boards : Number crunching : Panic Mode On (110) Server Problems?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.