climateprediction.net home page
How many hours are in a day?

How many hours are in a day?

Message boards : Number crunching : How many hours are in a day?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile old_user733
Avatar

Send message
Joined: 9 Aug 04
Posts: 25
Credit: 4,756,979
RAC: 0
Message 40155 - Posted: 16 Jul 2010, 12:56:45 UTC

Well, I wasn't actually saying that, but yes, I tried it and that is exactly what is happening. Here's the message log from the attempt:


7/16/2010 8:48:45 AM CPDN Beta work fetch suspended by user
7/16/2010 8:48:49 AM CPDN Beta suspended by user
7/16/2010 8:49:39 AM climateprediction.net update requested by user
7/16/2010 8:49:42 AM climateprediction.net Sending scheduler request: Requested by user.
7/16/2010 8:49:42 AM climateprediction.net Requesting new tasks
7/16/2010 8:49:44 AM climateprediction.net Scheduler request completed: got 0 new tasks
7/16/2010 8:49:44 AM climateprediction.net Message from server: No work sent
7/16/2010 8:49:44 AM climateprediction.net Message from server: (reached daily quota of 4 tasks)

In the meantime, I still have the three tasks running, and 5 cores idle, but I will un-suspend Beta and they will get busy.

Also, every time I force CPDN to update and it gets this quota message, it backs off communication for a long time, in this case 11 hours. I then have to force it again when it's waiting to do a trickle.
ID: 40155 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 40156 - Posted: 16 Jul 2010, 13:45:46 UTC

As soon as you suspended the computer's Beta models, Boinc should have detected the 5 idle cores and requested at least 5 new CPDN models. AFAIK, detection of the idle cores should have made Boinc override the backoff and immediately ask itself without you needing to update CPDN.

I can't test the CPDN work fetch myself because both my computers are way over the 10-day buffer limit for CPDN.

I think the only daily quota value that now means anything with regard to work fetch is 0.


Cpdn news
ID: 40156 · Report as offensive     Reply Quote
Profile old_user733
Avatar

Send message
Joined: 9 Aug 04
Posts: 25
Credit: 4,756,979
RAC: 0
Message 40159 - Posted: 16 Jul 2010, 17:21:41 UTC

Well, nothing new here. I've been running BOINC 6.10.56 since it came out. The problem started happening several weeks ago, and on both of my machines, though somehow the other one managed to get a couple of tasks, even though it was also getting the quota message.

Getting the quota message apparently tells BOINC to back off on communications, including trickles, so I now find myself checking every so often and doing them manually. If BOINC then asks for tasks again, it again gets the quota message from the server. Just did it again:


7/16/2010 1:09:38 PM climateprediction.net update requested by user
7/16/2010 1:09:41 PM climateprediction.net Sending scheduler request: Requested by user.
7/16/2010 1:09:41 PM climateprediction.net Requesting new tasks
7/16/2010 1:09:43 PM climateprediction.net Scheduler request completed: got 0 new tasks
7/16/2010 1:09:43 PM climateprediction.net Message from server: No work sent
7/16/2010 1:09:43 PM climateprediction.net Message from server: (reached daily quota of 4 tasks)

And again, BOINC has set a communications backoff for 6-1/2 hours this time. Mind you, I haven't had any new work from the project since June 29th, so I don't know where this quota is getting triggered.

What I was thinking is that it might have to do with the 20 tasks the server thinks that I have (which apparently were lost in the transactions), but even then, I haven't gotten anything recently, so why the "daily quota of 4"?

I guess I can wait until the work I have is finished, then detach and reattach to the project. That may fix it for my system, at least temporarily, but won't fix the real problem, which has got to be on the server side.
ID: 40159 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 40160 - Posted: 16 Jul 2010, 17:41:39 UTC
Last modified: 16 Jul 2010, 17:45:05 UTC

You'll see from my long post of 26 June in the thread for reporting model-wasting computers that the problem with Milo's quota-minussing script started after the server's Boinc upgrade which was on 3 June. This is why I've stopped reporting computers wasting multiple models. The quota-minussing script will only work for some computers. The daily quota isn't under control for all computers.
Cpdn news
ID: 40160 · Report as offensive     Reply Quote
Profile old_user733
Avatar

Send message
Joined: 9 Aug 04
Posts: 25
Credit: 4,756,979
RAC: 0
Message 40161 - Posted: 16 Jul 2010, 19:23:57 UTC

Hi Mo,

Yes I read that thread, and was lamenting that so many people wouldn't notice that their machines were erroring out everything they got sent, and I also noticed your post alluding to phantom tasks on Jan 27 and again on Mar 12, which is what I think is happening here. These tasks never showed up on my system, which, AFAIK, does not have any networking issues. If they had showed up and then gotten errors, I would have taken appropriate action.

BTW, when your quota is "minussed", how does it show in the message from the server? Does it actually say -1?

-Gene
ID: 40161 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 40162 - Posted: 16 Jul 2010, 21:48:16 UTC

One of the minussed computers can be seen here. I've no idea whether that -1 will ever show up in Boinc manager messages. It may well appear if Boinc keeps trying to fetch more models. The email definitely informs the owners about the minus quota.

But only about 1 in 15 of the people who receive the email post in the thread they're given a link to. These are for the most part unattended and unmonitored computers and I can assure you that even the owners who take up the invitation to post are singularly uncommunicative. The last thing they want is to have extended chats about their Boinc manager messages.

This is how Boinc's supposed to work. Set and forget. But of course if it wasn't for the small proportion of members who do post we'd never get to the root of many problems.

BTW, Thyme Lawn says that since the server's Boinc upgrade the only daily quota value that now means anything is -1. Not 0 as I said earlier.
Cpdn news
ID: 40162 · Report as offensive     Reply Quote
Profile [B@H] Ray
Avatar

Send message
Joined: 19 Aug 05
Posts: 104
Credit: 1,866,495
RAC: 0
Message 40163 - Posted: 16 Jul 2010, 22:29:36 UTC - in response to Message 40150.  

Lots of people are getting that quota message. Usually if you wait a while (several hours) it will give you a replacement WU.

Jim
I am glad that worked for you and some others, many of us are still getting that message. My last two request are here.

7/14/2010 12:48:07 PM climateprediction.net work fetch resumed by user
7/14/2010 12:48:11 PM climateprediction.net Sending scheduler request: To fetch work.
7/14/2010 12:48:11 PM climateprediction.net Requesting new tasks
7/14/2010 12:48:13 PM climateprediction.net Scheduler request completed: got 0 new tasks
7/14/2010 12:48:13 PM climateprediction.net Message from server: No work sent
7/14/2010 12:48:13 PM climateprediction.net Message from server: (reached daily quota of 1 tasks)

7/16/2010 6:13:32 PM climateprediction.net work fetch resumed by user
7/16/2010 6:13:36 PM climateprediction.net Sending scheduler request: To fetch work.
7/16/2010 6:13:36 PM climateprediction.net Requesting new tasks
7/16/2010 6:13:37 PM climateprediction.net Scheduler request completed: got 0 new tasks
7/16/2010 6:13:37 PM climateprediction.net Message from server: No work sent
7/16/2010 6:13:37 PM climateprediction.net Message from server: (reached daily quota of 1 tasks)

There are over two days apart.

Just hope that when the one running completes I can get another at that time.
ID: 40163 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 40164 - Posted: 16 Jul 2010, 23:56:22 UTC - in response to Message 40163.  

Ray
When the server says that your quota is 1, what does the web site say for that computer?


Backups: Here
ID: 40164 · Report as offensive     Reply Quote
Profile old_user733
Avatar

Send message
Joined: 9 Aug 04
Posts: 25
Credit: 4,756,979
RAC: 0
Message 40165 - Posted: 17 Jul 2010, 3:08:10 UTC

Whoa. Suddenly got some new tasks for machine 1070830. 5 in all, so now all 8 cores are busy with these making up the debt, and the Beta tasks are in waiting. From looking at the task list, only these 5 have been added, no new phantoms. So, problem solved for the time being, at least for this machine. Thanks all.

ID: 40165 · Report as offensive     Reply Quote
Profile [B@H] Ray
Avatar

Send message
Joined: 19 Aug 05
Posts: 104
Credit: 1,866,495
RAC: 0
Message 40166 - Posted: 17 Jul 2010, 5:55:10 UTC - in response to Message 40164.  
Last modified: 17 Jul 2010, 6:03:28 UTC

Ray
When the server says that your quota is 1, what does the web site say for that computer?


The web site says:
Maximum daily WU quota per CPU 4/day

But with the new server scripts that means nothing as it has separete quota's for each type of model.

As I do have one I will just run that one and get a new one (I hope) when it is completed. In a while VIP will have work again and I will just run that on this system till they run out than be back to CPDN. The other two systems will keep running CPDN along with VIP when VIP has work. For now it is happy running one model and Prime Grid. I just hope that I can get another model when the one running now completes, will have to wait a few days to find out it I can. This started having problems when I had the two models with "Error while downloading", before that I had no problems here. It probably is something buried deep in the code and as long as others can get work I will not worry about it. I do have Result 11555621 running good now.

The system giving problems is a AMD Athlon II 235e, Computer 1053982. My Intel systems are all good here.
ID: 40166 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 40168 - Posted: 18 Jul 2010, 20:59:08 UTC - in response to Message 40166.  

Ray
What you have seen is the same as what I had a couple of weeks back. (Web account/manager messages)

My theory on this is that the number shown in "daily quota" is a left over from what ever was there before the upgrade. A bit like the values shown for credits in "claimed" and "pending" - BOINC writes numbers there when a model finishes, and within 24 hours the project's credit script overwrites the "granted" column with a new figure, and may or may not clear the other 2 columns.

With the quota, there are 6 applications listed, so there should be 6 per core per day shown, which isn't the case.
So the real value(s) is(are) in a variable(s) which we can't see.
Until Milo can come up with a solution, (which may require the BOINC developers to fix it with a new version asap), crunchers here are stuck with waiting patiently for various hidden counters to time out, and allow more models to be sent. This seems to be happening after 2-3 days at present.


Backups: Here
ID: 40168 · Report as offensive     Reply Quote
Holmis

Send message
Joined: 31 Aug 04
Posts: 17
Credit: 303,467
RAC: 0
Message 40183 - Posted: 19 Jul 2010, 21:47:30 UTC

I don't know if this could be of help but on the Seti project they have enabled a new page under every participants computer details called "Application details".
This page lists information on which applications have been run on the host and what quotas are in affect.

I don't know how to enable this page or if the servers are running the same version of the software.

For an example this is how it looks. And this is the computer details page. I just picked some one because I haven't run Seti as of late.

If this could be enabled here it might show some light one the problems at hand.
ID: 40183 · Report as offensive     Reply Quote
Profile mo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 40185 - Posted: 20 Jul 2010, 2:44:27 UTC

Thanks for those interesting links. Our CPDN computer details pages certainly have no link to an applications page.

I don't think CPDN has the same Boinc server version as Seti. The CPDN version hides the real quotas per application from the user. This is why the quota we see is invalid and irrelevant. The extra code to make the quota per application visible to the user was added later.

I see that my computer details pages for Seti and Aqua show links to an applications page. But my computer page in my Einstein account has no Applications link although Einstein also upgraded its Boinc server version recently.
Cpdn news
ID: 40185 · Report as offensive     Reply Quote
Profile [B@H] Ray
Avatar

Send message
Joined: 19 Aug 05
Posts: 104
Credit: 1,866,495
RAC: 0
Message 40187 - Posted: 20 Jul 2010, 12:59:46 UTC

One model completed today, and the computer downloaded two right after that. Guess that the server just wanted to be sure mine could complete one before resetting the quota after those download errors. Hope others with this can get them after reporting.
Keep on crunching Pizza@Home
ID: 40187 · Report as offensive     Reply Quote
Profile JIM

Send message
Joined: 31 Dec 07
Posts: 1152
Credit: 22,053,321
RAC: 4,417
Message 40201 - Posted: 22 Jul 2010, 1:27:41 UTC

I was just wondering, is it possible to go back to the old Boinc server software. It worked a whole lot better. ;)

ID: 40201 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 40217 - Posted: 22 Jul 2010, 21:55:39 UTC - in response to Message 40201.  

Unlikely.
The server version was way behind, and several features were needed, such as "Plan class", which will allow the server to allocate model types only to computers that can handle them.

An upgrade has been tried several times in the past year, but hasn't worked for various reasons, and time has been short. The 'trigger' for a concerted effort this time apparently, is to test the different way of doing things. This was first run on SETI, (which is used as a test project for these things), and now here, because one of the changes is a radically different way of dealing with credits. Not used yet, I think, but in the future ...

We'll just have to make do, until it gets sorted.
Remember, the code has to be re-compiled each time a change is made, and this can take hours, so it's not something to rush into each time someone has a idea. :)


Backups: Here
ID: 40217 · Report as offensive     Reply Quote
Profile [B@H] Ray
Avatar

Send message
Joined: 19 Aug 05
Posts: 104
Credit: 1,866,495
RAC: 0
Message 40256 - Posted: 28 Jul 2010, 18:19:45 UTC

The system that was giving problems is doing good now. Another stated having problems today when it reported a complete model.

7/28/2010 4:14:57 AM climateprediction.net Computation for task famous_umdc_1399_200_006661835_5 finished
7/28/2010 4:18:12 AM climateprediction.net Finished upload of famous_umdc_1399_200_006661835_5_20.zip
7/28/2010 7:04:28 AM climateprediction.net Reporting 1 completed tasks, not requesting new tasks
7/28/2010 7:04:33 AM climateprediction.net Scheduler request completed
7/28/2010 9:55:16 AM climateprediction.net Sending scheduler request: To fetch work.
7/28/2010 9:55:16 AM climateprediction.net Requesting new tasks
7/28/2010 9:55:21 AM climateprediction.net Scheduler request completed: got 0 new tasks
7/28/2010 9:55:21 AM climateprediction.net Message from server: No work sent
7/28/2010 9:55:21 AM climateprediction.net Message from server: (reached daily quota of 4 tasks)

As that system still has one running I will not worry about it. I have CPDN set to NNW and will try when the one running gets closer to the end.

The server did issue a Phantem unit to it this morning, but I only recieved the reached daily quota of 4 tasks message when that was sent.

Will try again when the running model completes. Will be good when Berkley finds what is causing this so all projects can benefit from the fix.
Keep on crunching Pizza@Home
ID: 40256 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 40342 - Posted: 9 Aug 2010, 22:11:19 UTC

Milo has done a new version compile of a couple of daemons to do with downloading, so please see if you can get new work now.

ID: 40342 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : How many hours are in a day?

©2024 climateprediction.net