Just got work....

---

I'm getting it sporadically, but there are frequent periods where I'm getting nothing.

-Dave

---

beem getting stuttering batches with two of my guys sitting idle now for last half hr

Looks like something weird is going on with the server. There are about 18000 workunits ready to be sent out (which is quite a lot).

--Travis

What happened to all those???

I see my unattended MW machine has only 2 tasks left, there must be an task issue again :( :( :( :( :( :( :( :( Only 55 Ones ready to send. Seems machine dont get anyones @ the moment..

---

DSKAG Austria Research Team: http://www.research.dskag.at

10/25/2011 1:19:26 PM Milkyway@home Requesting new tasks for GPU
10/25/2011 1:19:31 PM Milkyway@home Scheduler request completed: got 0 new tasks
10/25/2011 1:19:31 PM Milkyway@home Message from server: No work available

---

Hi Guys,

The problem should be fixed -- I put in some changes yesterday which limits the maximum amount of workunits that can be in the database at any given time; since the server had around 300,000 WUs in the database when i made the change I put the cap at 400,000. Looks like you guys requested a ton more work and there are 400,000 WUs in the database at the moment. I'm going to increase the cap to 500,000 which hopefully should be more than enough.

The reason I'm putting the cap in is to prevent the problem we had earlier when I had to wipe clean the database. For whatever reason sometimes when the database is being backed up to disk or if there's a problem with the feeder, the code that calculates the number of workunits available fails and the work generator continues to generate new workunits without stopping -- flooding the database. This extra check should prevent that from happening.

Anyways, work should be flowing again shortly; and I'm working on a better solution to this problem then just a hard cap on the number of workunits in the database (although we probably need that to keep the database from getting unresponsive).

--Travis

---

A Big Thanks Travis!

All my hungry crunchers are happy again!!

We seem to be back to having no new tasks available again...

We seem to be back to having no new tasks available again...

I saw that, but it seems to be working ATM...

10/26/2011 11:14:14 AM Milkyway@home Scheduler request completed: got 2 new tasks
10/26/2011 11:15:24 AM Milkyway@home Scheduler request completed: got 1 new tasks
10/26/2011 11:16:33 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
10/26/2011 11:17:43 AM Milkyway@home Scheduler request completed: got 4 new tasks
10/26/2011 11:18:50 AM Milkyway@home Scheduler request completed: got 1 new tasks

---

We seem to be back to having no new tasks available again...

I saw that, but it seems to be working ATM...

10/26/2011 11:14:14 AM Milkyway@home Scheduler request completed: got 2 new tasks
10/26/2011 11:15:24 AM Milkyway@home Scheduler request completed: got 1 new tasks
10/26/2011 11:16:33 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
10/26/2011 11:17:43 AM Milkyway@home Scheduler request completed: got 4 new tasks
10/26/2011 11:18:50 AM Milkyway@home Scheduler request completed: got 1 new tasks

It's just that we're up against the limit again. :P I'll bump it up to 600,000 workunits and see if the database survives.

---

Would it help to stop some of the computers putting out continual error's or hogging hundreds of wus that don't get computed? Taking care of those should help to bring the number down.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Would it help to stop some of the computers putting out continual error's or hogging hundreds of wus that don't get computed? Taking care of those should help to bring the number down.

The continual errors certainly would, but AFAIK there's no way we can stop those machines from continually sending us junk. Is it possible to ban a host?

Not quite sure what to do about the WU hogs either, seems like that's a BOINC bug. Actually maybe I'll complain on the BOINC lists about that. Have a link to any of those offenders I could forward on?

---

[quote]Would it help to stop some of the computers putting out continual error's or hogging hundreds of wus that don't get computed? Taking care of those should help to bring the number down.

The continual errors certainly would, but AFAIK there's no way we can stop those machines from continually sending us junk. Is it possible to ban a host?

Yes. It was done on Nanohive years ago. Don't know how but he did ban some hosts or users. IP addy maybe?

Not quite sure what to do about the WU hogs either, seems like that's a BOINC bug. Actually maybe I'll complain on the BOINC lists about that. Have a link to any of those offenders I could forward on?

A few threads exist with both listes (errors and timeouts). Some may have been taken care of, I'm not going to check all of the hosts right now.
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2538
http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=2180#45752

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Perhaps a normal Firewall deny rule for a specified IP of the errorous hosts would help.

---

DSKAG Austria Research Team: http://www.research.dskag.at

10/30/2011 5:00:48 AM Milkyway@home Scheduler request completed: got 0 new tasks
10/30/2011 5:00:48 AM Milkyway@home Message from server: Project has no tasks available
10/30/2011 5:01:58 AM Milkyway@home Message from server: No work available
10/30/2011 5:04:05 AM Milkyway@home Message from server: No work available
10/30/2011 5:07:22 AM Milkyway@home Message from server: No work available
10/30/2011 5:08:29 AM Milkyway@home Message from server: No work available
10/30/2011 5:15:05 AM Milkyway@home Message from server: No work available
10/30/2011 5:20:52 AM Milkyway@home Message from server: Project has no tasks available
10/30/2011 5:25:57 AM Milkyway@home Message from server: No work available
10/30/2011 5:31:15 AM Milkyway@home Scheduler request completed: got 15 new tasks
10/30/2011 5:32:24 AM Milkyway@home Scheduler request completed: got 9 new tasks

---

The continual errors certainly would, but AFAIK there's no way we can stop those machines from continually sending us junk. Is it possible to ban a host?

Would be worth having a chat with GPUGRID admins - they used a choke mechanism a while back, AFAIK its still there - to stop runaway hosts. Its a mechanism inside the server software based on counting bad WUs over a set period of time, then gradually increasing the amount they can download as reliability returns. Both the choke and rate of returning to "normal" are configurable in the server software.

Worth checking out to see if its applicable at MW.

Maybe these run away hosts belong to farms. Some have quite a few machines running on full auto and never really checked. Why not just send them a PM?

Just in time....

11/19/2011 4:46:29 AM Milkyway@home Reporting 2 completed tasks, requesting new tasks for GPU
11/19/2011 4:46:31 AM Milkyway@home Scheduler request completed: got 0 new tasks
11/19/2011 4:46:31 AM Milkyway@home Message from server: No tasks sent
11/19/2011 4:46:31 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
11/19/2011 4:47:37 AM Milkyway@home Reporting 2 completed tasks, requesting new tasks for GPU
11/19/2011 4:47:39 AM Milkyway@home Scheduler request completed: got 0 new tasks
11/19/2011 4:47:39 AM Milkyway@home Message from server: No tasks sent
11/19/2011 4:47:39 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
11/19/2011 4:48:44 AM Milkyway@home Reporting 1 completed tasks, requesting new tasks for GPU
11/19/2011 4:48:46 AM Milkyway@home Message from server: No tasks sent
11/19/2011 4:48:46 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
11/19/2011 4:49:52 AM Milkyway@home Scheduler request completed: got 0 new tasks
11/19/2011 4:49:52 AM Milkyway@home Message from server: No tasks sent
11/19/2011 4:49:52 AM Milkyway@home Message from server: Tasks for CPU are available, but your preferences are set to not accept them
11/19/2011 4:50:57 AM Milkyway@home Reporting 2 completed tasks, requesting new tasks for GPU
11/19/2011 4:51:09 AM Milkyway@home Scheduler request completed: got 22 new tasks

---

Back up....

11/20/2011 4:44:36 AM Project communication failed: attempting access to reference site
11/20/2011 4:44:36 AM Milkyway@home Scheduler request failed: Timeout was reached
11/20/2011 4:44:37 AM Internet access OK - project servers may be temporarily down.
11/20/2011 4:45:36 AM Milkyway@home Reporting 13 completed tasks, requesting new tasks for GPU
11/20/2011 4:46:19 AM Project communication failed: attempting access to reference site
11/20/2011 4:46:19 AM Milkyway@home Scheduler request failed: Failure when receiving data from the peer
11/20/2011 4:46:20 AM Internet access OK - project servers may be temporarily down.
11/20/2011 4:47:19 AM Milkyway@home Sending scheduler request: To fetch work.
11/20/2011 4:47:19 AM Milkyway@home Reporting 16 completed tasks, requesting new tasks for GPU
11/20/2011 4:48:02 AM Project communication failed: attempting access to reference site
11/20/2011 4:48:02 AM Milkyway@home Scheduler request failed: Failure when receiving data from the peer
11/20/2011 4:48:03 AM Internet access OK - project servers may be temporarily down.
11/20/2011 4:49:02 AM Milkyway@home Sending scheduler request: To fetch work.
11/20/2011 4:49:02 AM Milkyway@home Reporting 19 completed tasks, requesting new tasks for GPU
11/20/2011 4:54:31 AM Project communication failed: attempting access to reference site
11/20/2011 4:54:31 AM Milkyway@home Scheduler request failed: Timeout was reached
11/20/2011 4:54:32 AM Internet access OK - project servers may be temporarily down.
11/20/2011 4:55:33 AM Milkyway@home Sending scheduler request: To fetch work.
11/20/2011 4:55:33 AM Milkyway@home Reporting 22 completed tasks, requesting new tasks for GPU
11/20/2011 4:56:16 AM Project communication failed: attempting access to reference site
11/20/2011 4:56:16 AM Milkyway@home Scheduler request failed: Failure when receiving data from the peer
11/20/2011 4:56:18 AM Internet access OK - project servers may be temporarily down.
11/20/2011 4:57:26 AM Milkyway@home Reporting 22 completed tasks, requesting new tasks for GPU
11/20/2011 4:58:09 AM Milkyway@home Scheduler request failed: Failure when receiving data from the peer
11/20/2011 4:58:11 AM Internet access OK - project servers may be temporarily down.
11/20/2011 5:04:04 AM Milkyway@home Sending scheduler request: To fetch work.
11/20/2011 5:04:04 AM Milkyway@home Reporting 22 completed tasks, requesting new tasks for GPU
11/20/2011 5:09:11 AM Project communication failed: attempting access to reference site
11/20/2011 5:09:11 AM Milkyway@home Scheduler request failed: Timeout was reached
11/20/2011 5:09:12 AM Internet access OK - project servers may be temporarily down.
11/20/2011 5:26:00 AM Milkyway@home Sending scheduler request: To fetch work.
11/20/2011 5:26:00 AM Milkyway@home Reporting 22 completed tasks, requesting new tasks for GPU
11/20/2011 5:26:06 AM Milkyway@home Scheduler request completed: got 23 new tasks

---

I guess it was a hick-up as I am still getting new WU's for the ATI-app.

---

Greetings from,
TJ