Ghost WU issue (and some talk about deadlines)

Message boards : Number crunching : Ghost WU issue (and some talk about deadlines)
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 5 · 6 · 7 · 8 · 9 · 10 · 11 . . . 12 · Next

AuthorMessage
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571491 - Posted: 19 May 2007, 18:27:21 UTC - in response to Message 571472.  

I don't know if others are experiencing this. But after renaming the app_info.xml file I was able to download and upload without renaming it back. I'm running optimized and the optimized client continues to run regardless of the app_info.xml file being present or not.

WU's are authenticating and all seem to be working fine for now.


I ran into this when I ran optimized over on Beta by mistake (don't do it!) and removed to app info file to go back to normal processing.
What happens is that the WUs downloaded when the app info file was in place will run with the optimized app, but any new work will revert back to the stock app.


Actually, right now I'm not sure how this works. I had a rig this morning that had no Seti work. Took out the app info file, got some WUs. Put the app info file back and started crunching. Took the app info file back out, and it looks like it's still using the optimized app.
Please note that each time, Boinc was stopped and restarted.
When you start Boinc with the xml file in place, does it then tag all work in the cache to run with the optimized app, or just the ones it happens to be processing at the moment? If all the WUs get tagged, you could remove the app info xml file, get work, replace the file, start crunching, and remove it again to allow comms with the host as usual.
What I'm not clear on is when Boinc will decide to use the stock app instead of the optimized one.



I honestly don't know what's going on, only that additional wu's have been download since yesterday. And with the app_info.xml file still renamed as app_info.xml_rename, crunching continues with the newly downloaded wu's with the optimized client. I've renamed the stock app so it's not only not running. Boinc has not downloaded a new one either.

So I don't get it.....


I posted a question about this for Simon over on his site. Hopefully he or Joe will have the answers and will report back to us here.
"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571491 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571569 - Posted: 19 May 2007, 19:22:18 UTC - in response to Message 571479.  

...
What I'm not clear on is when Boinc will decide to use the stock app instead of the optimized one.


The Apps seems to be mentioned in the client_state.xml, somehow , in my case, KWSN_2.2B_SSE2-P4_Ben-Joe.exe has inserted itself where the standard app used to be... worth a closer look



I also seem to recall that the first time I restarted Boinc without the app info file, some of the rigs went through some sort of update, downoading Seti program files, and I don't think some of them did that.
This might play into the whole scenario as well.
"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571569 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571572 - Posted: 19 May 2007, 19:24:38 UTC - in response to Message 571569.  
Last modified: 19 May 2007, 19:25:10 UTC


...
This might play into the whole scenario as well.


Of course I'd rather they'd just unbreak the scheduler code ..... than all this muckin' 'bout :D

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571572 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571588 - Posted: 19 May 2007, 19:44:53 UTC

And here's another update....
Just checked one of my rigs and saw in the messages that it was hungry and trying to get new work.
Renamed the app info file and restarted Boinc. It connected to Seti and downloaded 20 new WUs. Went into EDF mode because they were small WUs, so it suspended processing the 2 it was doing and started 2 of the brand new WUs. Checked task manager, and the new WUs are using the 2.2b optimized app! And this after restart of Boinc without the app info file active.
Not complaining! But it would be informative as to how this mechanism of optimized app selection is really working.
If whatever selects the optimized app is present in another file (one of the state files?) even without the app info file present, then we can avoid the current problem just by inactivating the app info file and leave it inactive until that other file gets updated (by what, I don't know).
'Tis still a puzzlement.
"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571588 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571591 - Posted: 19 May 2007, 19:50:20 UTC - in response to Message 571588.  

...
'Tis still a puzzlement.


I have a feeling if you happened to let the cache run dry, with the appinfo disabled, then the standard app may download and install... if that happens then the three chicken files would have to be reinstalled.

Just a thought...

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571591 · Report as offensive
crazyrabbit1

Send message
Joined: 17 Sep 06
Posts: 35
Credit: 2,282,319
RAC: 0
Germany
Message 571594 - Posted: 19 May 2007, 19:55:15 UTC

Just a thought from a guy that know nothing about software (me). What can happen if we rename the chicken app to the name of the original app and run without the appinfo. Is this possible??
ID: 571594 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 571608 - Posted: 19 May 2007, 20:10:50 UTC

Let me throw another twist into this twisted situation.

I was able to get a couple of WU's with the renaming process but for the life of me I couldn't get it to crunch them with chicken soup.

So, I decided to shut down my computer to clean up all the computer memory, cashe, etc.

I restarted the 'puter and SETI and low and behold, it was crunchin on chicken soup now (checked via task mgr)

Maybe it will work for others.

ID: 571608 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571614 - Posted: 19 May 2007, 20:18:03 UTC - in response to Message 571608.  
Last modified: 19 May 2007, 20:18:54 UTC

LOL: I think I'm starting to get it, We are trying to reason about a situation that has arisen from what was a completely illogical fault in the first place. Why is (or should) the scheduler be treating 'anonymous platform' uploads and downloads any different to the 'stock platform'. I think that while that bug remains a mystery, the remaining bizarre function of the mechanism may too.
[ Kind of like how "A stopped Clock Tells the Right Time Twice a Day" ]

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571614 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 571618 - Posted: 19 May 2007, 20:20:52 UTC - in response to Message 571614.  

LOL: I think I'm starting to get it, We are trying to reason about a situation that has arisen from what was a completely illogical fault in the first place. Why is (or should) the scheduler be treating 'anonymous platform' uploads and downloads any different to the 'stock platform'. I think that while that bug remains a mystery, the remaining bizarre function of the mechanism may too.
[ Kind of like how "A stopped Clock Tells the Right Time Twice a Day" ]


DO do DO do, You are now entering the Twilight Zone.
ID: 571618 · Report as offensive
James Nelson
Volunteer tester
Avatar

Send message
Joined: 23 Mar 02
Posts: 381
Credit: 4,806,382
RAC: 0
United States
Message 571619 - Posted: 19 May 2007, 20:21:17 UTC - in response to Message 571594.  

Just a thought from a guy that know nothing about software (me). What can happen if we rename the chicken app to the name of the original app and run without the appinfo. Is this possible??

I think that the client checks the app against a checksum value and will think the app has become corupt because the optomised app will return the wrong value.
if someone else knows for sure let us know.
ID: 571619 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571622 - Posted: 19 May 2007, 20:23:46 UTC - in response to Message 571608.  

Let me throw another twist into this twisted situation.

I was able to get a couple of WU's with the renaming process but for the life of me I couldn't get it to crunch them with chicken soup.

So, I decided to shut down my computer to clean up all the computer memory, cashe, etc.

I restarted the 'puter and SETI and low and behold, it was crunchin on chicken soup now (checked via task mgr)

Maybe it will work for others.


The fact that you could not crunch with Chicken once you got the WUs would make one think that perhaps it was not active (did you have any Seti WUs left?) and that may be how you got the WUs without doing the workaround in the first place. When you rebooted the computer and Boinc restarted, it used the app info file still in place and reverted to Chicken again.
This may be yet another clue in this twisted puzzle. Joe Segur posted in the Beta NC forum something about the optimized app being linked to Boinc even without the app info file as long as there were some WUs still being crunched using the optimized app. Once there is no work using it, it is suspected that the link goes away and then the stock app takes over.

"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571622 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571624 - Posted: 19 May 2007, 20:25:24 UTC - in response to Message 571618.  


DO do DO do, You are now entering the Twilight Zone.


Let's figure out how many more random variables we can throw into the mix... that still make it work but can not be deciphered and understood :D

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571624 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571630 - Posted: 19 May 2007, 20:28:26 UTC - in response to Message 571622.  


The fact that you could not crunch with Chicken once you got the WUs would make one think that perhaps it was not active (did you have any Seti WUs left?) and that may be how you got the WUs without doing the workaround in the first place. When you rebooted the computer and Boinc restarted, it used the app info file still in place and reverted to Chicken again.
This may be yet another clue in this twisted puzzle. Joe Segur posted in the Beta NC forum something about the optimized app being linked to Boinc even without the app info file as long as there were some WUs still being crunched using the optimized app. Once there is no work using it, it is suspected that the link goes away and then the stock app takes over.


That sounds closer.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571630 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571633 - Posted: 19 May 2007, 20:30:51 UTC - in response to Message 571624.  
Last modified: 19 May 2007, 20:31:41 UTC


DO do DO do, You are now entering the Twilight Zone.


Let's figure out how many more random variables we can throw into the mix... that still make it work but can not be deciphered and understood :D


Matt should have quite a time sorting through all these posts when he gets back from vacation and starts to straighten the whole mess out.
I hope he takes the time to do so, I think we have reported some valuable clues as to what is going wrong and how certain work-arounds are behaving. Should make more sense to someone who knows more about the 'big picture' of how the pieces of this whole Seti crunching machine fits together and function.

"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571633 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 571635 - Posted: 19 May 2007, 20:31:27 UTC - in response to Message 571622.  
Last modified: 19 May 2007, 20:38:05 UTC

Let me throw another twist into this twisted situation.

I was able to get a couple of WU's with the renaming process but for the life of me I couldn't get it to crunch them with chicken soup.

So, I decided to shut down my computer to clean up all the computer memory, cashe, etc.

I restarted the 'puter and SETI and low and behold, it was crunchin on chicken soup now (checked via task mgr)

Maybe it will work for others.


The fact that you could not crunch with Chicken once you got the WUs would make one think that perhaps it was not active (did you have any Seti WUs left?) and that may be how you got the WUs without doing the workaround in the first place. When you rebooted the computer and Boinc restarted, it used the app info file still in place and reverted to Chicken again.


[EDIT] I did the work around to get the 2 WU's [END OF EDIT]

I didn't have any WU's at the time I got my 2 new WU's. They were the first ones I have received in xx days/weeks. So "I guess" that since the app file had not ben renamed back yet (cause it starts crunching right away??) it started running on the stock app. When I turned it off and back on, the file had been renamed back by that time and memory/cashe cleared.

I have no idea, help, I'm getting a migrane, lol.


This may be yet another clue in this twisted puzzle. Joe Segur posted in the Beta NC forum something about the optimized app being linked to Boinc even without the app info file as long as there were some WUs still being crunched using the optimized app. Once there is no work using it, it is suspected that the link goes away and then the stock app takes over.


Help, I've fallen and can't get back up.

[EDIT] I did this process on both 'puters and the situation was exactly the same with the exact same result.[EOE]


ID: 571635 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571643 - Posted: 19 May 2007, 20:36:18 UTC - in response to Message 571633.  


Matt should have quite a time sorting through all these posts when he gets back from vacation and starts to straighten the whole mess out.
I hope he takes the time to do so, I think we have reported some valuable clues as to what is going wrong and how certain work-arounds are behaving. Should make more sense to someone who knows more about the 'big picture' of how the pieces of this whole Seti crunching machine fits together and function.


Yes, During and after the Thumper outage, I got the distinct impression of a convergence of long standing symptoms that before appeared separate. (difficulty getting work when there should be plenty, ghost results, validate errors and scheduler borking) It'll be very interesting.

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571643 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51583
Credit: 1,018,363,574
RAC: 1,004
United States
Message 571653 - Posted: 19 May 2007, 20:47:28 UTC - in response to Message 571643.  


Matt should have quite a time sorting through all these posts when he gets back from vacation and starts to straighten the whole mess out.
I hope he takes the time to do so, I think we have reported some valuable clues as to what is going wrong and how certain work-arounds are behaving. Should make more sense to someone who knows more about the 'big picture' of how the pieces of this whole Seti crunching machine fits together and function.


Yes, During and after the Thumper outage, I got the distinct impression of a convergence of long standing symptoms that before appeared separate. (difficulty getting work when there should be plenty, ghost results, validate errors and scheduler borking) It'll be very interesting.


Yes, I think it's been one of those 'perfect storm' scenarios.

"Time is simply the mechanism that keeps everything from happening all at once."

ID: 571653 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571658 - Posted: 19 May 2007, 20:49:21 UTC - in response to Message 571635.  
Last modified: 19 May 2007, 20:49:58 UTC


...
Help, I've fallen and can't get back up.
...


Actually the 'Anonymous Platform' fell out from under us ALL, we are in freefall, Grasping for straws which for some bizarre reason are taking quite a bit of our combined weight .... [for the moment]

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571658 · Report as offensive
Odysseus
Volunteer tester
Avatar

Send message
Joined: 26 Jul 99
Posts: 1808
Credit: 6,701,347
RAC: 6
Canada
Message 571718 - Posted: 19 May 2007, 21:31:07 UTC
Last modified: 19 May 2007, 21:32:14 UTC

FWIW here are a few observations of my experience with the app_info trick on my Mac G5, running BOINC v5.4.9:

This host runs the Alex Kan / Team MacNN optimized app, and had been generating “ghosts” every few minutes for a number of hours before I suspended the project. It had no S@h tasks on board (other than three Beta results that it’s been trying to report, without success, for some time).

I quit BOINC, renamed app_info.xml, relaunched, set S@h to NNT, Resumed and Updated, getting a successful not-requesting-or-reporting transaction on the first try.

I then Allowed new tasks: BOINC downloaded a bunch of work along with the stock v5.13 app and collateral. (The read-me file, of all things, stalled for a few minutes but eventually got through.) The app* began crunching briefly while I set S@h to NNT and quit BOINC again.

I reverted app_info.xml to its proper name and relaunched BOINC. It immediately deleted the stock app &c. from the project folder. It’s now happily crunching with the optimized app—but instead of displaying it by the name in app_info (viz v5.17, for compatibility with Beta), it continues to identify the cruncher as v5.13!

Uploads have proceeded OK so far, despite often taking several attempts before succeeding. Now we’ll see what happens when BOINC tries to report …

* To be honest, I’m not sure which one: I neglected to check it in Activity Monitor. BOINC was showing it as v5.13, but apparently that doesn’t necessarily mean anything.

ID: 571718 · Report as offensive
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 571720 - Posted: 19 May 2007, 21:33:52 UTC - in response to Message 571718.  

Crikey! it works on Macs :O

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 571720 · Report as offensive
Previous · 1 . . . 5 · 6 · 7 · 8 · 9 · 10 · 11 . . . 12 · Next

Message boards : Number crunching : Ghost WU issue (and some talk about deadlines)


 
©2026 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.