Panic Mode On (101) Server Problems?

Message boards : Number crunching : Panic Mode On (101) Server Problems?
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 15 · 16 · 17 · 18 · 19 · 20 · 21 . . . 27 · Next

AuthorMessage
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1744453 - Posted: 24 Nov 2015, 5:46:36 UTC - in response to Message 1744101.  

Splitter output even lower than it has been.
Grant
Darwin NT
ID: 1744453 · Report as offensive
Harri Liljeroos
Avatar

Send message
Joined: 29 May 99
Posts: 3989
Credit: 85,281,665
RAC: 126
Finland
Message 1744467 - Posted: 24 Nov 2015, 7:37:47 UTC

Is there a reason why several splitters are allowed to work on same "tape"? When this happens it slows down the creation of new tasks.
ID: 1744467 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1744474 - Posted: 24 Nov 2015, 8:49:53 UTC - in response to Message 1744467.  

Is there a reason why several splitters are allowed to work on same "tape"? When this happens it slows down the creation of new tasks.

It's been that way for several weeks now, although for whatever reason splitter output had been at their highest levels in years even with as many as 4 splitters on the one file.
Unfortunately for some reason they've reverted back to their low output mode since the last weekly outage.
Grant
Darwin NT
ID: 1744474 · 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: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1744495 - Posted: 24 Nov 2015, 11:45:52 UTC

...However as I type the splitters are running at ~37 WU/second, so Grant, as ever, is demonstrating he is a glass nearly empty person who does not understand how caches work.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1744495 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1744499 - Posted: 24 Nov 2015, 12:27:40 UTC - in response to Message 1744495.  

...However as I type the splitters are running at ~37 WU/second, so Grant, as ever, is demonstrating he is a glass nearly empty person who does not understand how caches work.

And 37 per second is 133,200 per hour - more than enough to replace the 110,956 being returned per hour. So we're building our own caches instead of RTS: checking mine, all my hosts which are configured to consume an entire quota in less than 24 hours are fully at the limit. No panic here.
ID: 1744499 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1744679 - Posted: 25 Nov 2015, 6:45:05 UTC - in response to Message 1744495.  

...However as I type the splitters are running at ~37 WU/second, so Grant, as ever, is demonstrating he is a glass nearly empty person who does not understand how caches work.

And Rob continues to display his considerable ignorance & continues to carp on about me making a simple, factual, statement as he is unable to read & understand graphs & how they relate to each other.
Very sad.
Grant
Darwin NT
ID: 1744679 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1744681 - Posted: 25 Nov 2015, 6:55:30 UTC - in response to Message 1744499.  

And 37 per second is 133,200 per hour - more than enough to replace the 110,956 being returned per hour.

Unfortunately as has been pointed out many times in the past that is just the splitter output at that point in time - an instantaneous value - at other points in time it's less than 20/s; which is only 72,000 per hour, which is no where near enough required to meet demand these days, let alone build up a buffer.
What matters is the average output.

Looking at the graphs (for those that are able to comprehend such things) for the last couple of weeks you can see that for a while there, the average (not instantaneous) output of the splitters was greater than the demand, so we ended up with a ready-to-send buffer.
For the week prior to this weeks outage you will see that the average splitter output was much reduced, resulting in only a few very brief periods where output exceed demand and a very small buffer was built up on couple of occasions. However overall, demand was greater than supply. So apart from those very few blips, the ready-to-send buffer didn't build up to it's limit.
Grant
Darwin NT
ID: 1744681 · 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: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1744695 - Posted: 25 Nov 2015, 8:12:23 UTC

Oh dear, what a shame....
The graphs show a different picture to the one portrayed by Grant - over the last year - with the exception of a slight dip in July & August the production rate has been sitting around 30/s. Indeed one can see the impact of the "great splitter debacle" of a couple of weeks ago - where the creation rate went sky-high, then collapsed, but of course that was a time when almost every WU split was rubbish...
Personally I would rather have a small RTS full of usable WU than a massive one full of rubbish...
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1744695 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1744704 - Posted: 25 Nov 2015, 9:06:32 UTC - in response to Message 1744695.  

Oh dear, what a shame....
The graphs show a different picture to the one portrayed by Grant - over the last year - with the exception of a slight dip in July & August the production rate has been sitting around 30/s. Indeed one can see the impact of the "great splitter debacle" of a couple of weeks ago - where the creation rate went sky-high, then collapsed, but of course that was a time when almost every WU split was rubbish...
Personally I would rather have a small RTS full of usable WU than a massive one full of rubbish...


I really hope you're taking the piss, because it would be sad if you really believe what you're saying.
Either way, it's a poor reflection on you.
Grant
Darwin NT
ID: 1744704 · Report as offensive
Profile William
Volunteer tester
Avatar

Send message
Joined: 14 Feb 13
Posts: 2037
Credit: 17,689,662
RAC: 0
Message 1744707 - Posted: 25 Nov 2015, 9:15:25 UTC

Can you cool it guys?
Just because you have different opinions as to whether the splitters are up to it or not you don't need to go and get nasty.

Besides I find the discussion rather academic.
If you want to get on a lengthy discussion about potential and real splitter output and filling levels I suggest you start a dedicated thread.
A person who won't read has no advantage over one who can't read. (Mark Twain)
ID: 1744707 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1744714 - Posted: 25 Nov 2015, 10:45:13 UTC

Ok guys, ten paces and then fire the links that you're both working from. :-D

Cheers.
ID: 1744714 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1744717 - Posted: 25 Nov 2015, 10:57:45 UTC - in response to Message 1744714.  

ID: 1744717 · Report as offensive
raydar115

Send message
Joined: 6 Oct 02
Posts: 17
Credit: 16,305,128
RAC: 0
United States
Message 1745363 - Posted: 28 Nov 2015, 9:01:58 UTC
Last modified: 28 Nov 2015, 9:05:29 UTC

YIKES WOULDN'T YOU KNOW ON A FRIDAY NIGHT THE Astropulse science database WENT OFF LINE JUST AS I NEEDED TO RELOAD LOL DANG GREMINS
ID: 1745363 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1745407 - Posted: 28 Nov 2015, 15:22:03 UTC
Last modified: 28 Nov 2015, 15:25:46 UTC

Can somebody abort all GPU AP's and all MB's for CPU please

I had to reload Lunatics and it abandoned a heap of units but the servers still think I'm doing them and so it will not send me any more just says I've hit my limit .

I also try'd to reset and all that did is abandon more units thank you

EDIT only on the AMD machine this machine http://setiathome.berkeley.edu/results.php?hostid=7486794
ID: 1745407 · 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: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1745436 - Posted: 28 Nov 2015, 17:31:36 UTC

You had a handful of tasks aborted a few weeks back, then today you had a pile of tasks timed out.....
Now you are left with a pile of ghosts.....
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1745436 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1745439 - Posted: 28 Nov 2015, 17:41:58 UTC - in response to Message 1745407.  

Can somebody abort all GPU AP's and all MB's for CPU please

I had to reload Lunatics and it abandoned a heap of units but the servers still think I'm doing them and so it will not send me any more just says I've hit my limit .

I also try'd to reset and all that did is abandon more units thank you

EDIT only on the AMD machine this machine http://setiathome.berkeley.edu/results.php?hostid=7486794

Detach and Reattach (Remove and Add project in New Boinc speak) your host, this will mark all those ghost tasks as abandoned.

Claggy
ID: 1745439 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1745603 - Posted: 29 Nov 2015, 12:56:23 UTC

Yes Rob I know that was possibly when I updated Lunatics but this time I was trying to change from the sxf 4.1 app to ssse3 and didn't remember I should have finished the units before changing the app. It was late and I was tired .

Amd does not like the sxf 4.1 app or what ever it's called

I then thought a reset mite fix it and lost all the AP's as well and that's when the penny dropped , "you iddiot glen of course it gona dump the units if there is no app for it to use , Then your a ____ head for doing a reset of the project ahhhhhh"

so that's how come I've lost this lot .

Dam Claggy I should have come here and checked to see if anybody has left any message I mite need .

I now have to many AP's to detach the project and do what you said .

So any chance some one can abort all units sent to me on the
should be 30 CPU MB's 6th November . Lost from upgrade Luna stuff up
101 sent on the 27th November CPU MB's .These are the SXF 4.1 units
126 sent on the 28th November GPU's MB's . These where the one lost on reset
120 AP's sent on 28th GPU's . These where lost on reset
2 AP's CPU sent on the 6th and
1 AP CPU on the 27th

The servers don't want to send me any more MB cpu units guess it still thinks I have my Quota with 101 lost . The ap's seem to be ok but splitters have stopped so no prob with them
ID: 1745603 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1745605 - Posted: 29 Nov 2015, 14:11:07 UTC - in response to Message 1745603.  

That analysis doesn't quite ring true. Tasks which have been 'discarded' - removed from BOINC Manager's display of tasks ready to run, because no application was found to process them - don't count towards the quota of tasks in progress: they have effectively become ghosts. There's no need for a member of staff to fly back from their home state on Thanksgiving weekend to manually remove those ghosts.

I see that you have received two new CPU tasks today - 4558304988 at 3:18:44 UTC, completed, and 4558847125 at 7:49:46 UTC, still in progress. So your CPU isn't quota limited - there should not be anything inhibiting CPU MB fetch, provided you currently have a viable application in place.

If you are still seeing 'reached a quota of tasks in progress' in your message log, they must be GPU tasks. Temporarily reduce your cache settings and burn off some GPU tasks - that should free things up.

Depending how many manual changes you've made to the app_info.xml file assembled by the Lunatics installer, there should be no need to empty the cache before switching from one application to another in the same category - say sse4.1 to ssse3: we supply the same <app_version> information for all apps. It is important, though, that you only try to run SIMD instruction sets that your CPU can physically run - check with CPU-Z if you're not sure. But I thought we'd cured that by adding CPU detection into the installer, and disabling the selection of incompatible options?
ID: 1745605 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1745607 - Posted: 29 Nov 2015, 14:12:37 UTC

Could some one tell me how you work out what the FFA values are , I'm getting this error for the GPU , GTX 650 1 gig ram 384 cuda's

Running on device number: 0
Sleep() & wait for event loops will be used in some places
DATA_CHUNK_UNROLL set to:12
oclFFT plan class overrides requested: global radix 256; local radix 16; max workgroup size 1024
WARNING: incorrect FFA thread fetch block override value:1024, using defaults
FFA thread block override value:2048

ID: 1745607 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1745609 - Posted: 29 Nov 2015, 14:24:57 UTC - in response to Message 1745607.  

Sorry guys looks like I figured it out error gone ....Stupid readme files confuse more than help :-)
ID: 1745609 · Report as offensive
Previous · 1 . . . 15 · 16 · 17 · 18 · 19 · 20 · 21 . . . 27 · Next

Message boards : Number crunching : Panic Mode On (101) 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.