Anyone notice a change with the new server settings?

---

It seems the same, I'll repost in the morning, wii have a better picture. Patrick

Not yet. But the 'Communication deferred' times seem more screwed up. I got 16 min, 7 min, 2 hours & 12 min, 42 min. Doesn't seem to have the 1 min backoff anymore (or atleast I'm not getting it at the moment).

*edit* nevermind on the 1 min backoff, got it finally.

---

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

Not yet. But the 'Communication deferred' times seem more screwed up. I got 16 min, 7 min, 2 hours & 12 min, 42 min. Doesn't seem to have the 1 min backoff anymore (or atleast I'm not getting it at the moment).

*edit* nevermind on the 1 min backoff, got it finally.

The server crashing probably didn't help for awhile there.

---

Well Travis, something definitely changed.
I got a series of messages that I have been getting for a while plus one message I don't recall seeing for a number of months now, they go like this

Sending scheduler request: To fetch work
Requesting 154355 seconds of new work
Scheduler RPC succeeded: [Server version 607]
Message from server: No work sent
Message from server: (Reached per-CPU limit of 6 tasks)
Message from server: (Project has no jobs available)
Deferring communication for 1 min 0 sec
Reason: No work from project

Have not seen 'Reached per-CPU limit of 6 tasks' for a very long time.

Unfortunately I have seen all the other messages heaps of times.

But it looks like an improvement, particularly if I can get the message a few more times.

Thanks Travis,

Conan.

---

02.06.2009 07:43:06 Milkyway@home Message from server: (Project has no jobs available)

It would be nice, if Travis could get the MW_GPU Server incl. GPU-WUs running, ATI first if nVIDIA isnt ready !

02.06.2009 07:43:06 Milkyway@home Message from server: (Project has no jobs available)

It would be nice, if Travis could get the MW_GPU Server incl. GPU-WUs running, ATI first if nVIDIA isnt ready !

I'm working on it. Hopefully WUs will be availalble at milkyway_gpu@home sometime this week.

---

It might be a good idea to get rid of the 1 min delay as even when your system waits 1 min. There is no wu to get. Its seems to be causing a bottle neck and not doing the job that it was intended to do.

Thanks


DD,

---

Unfortunately the 'Deferring Communication for 1min' isn't a forceful measure - it can still be overridden by clicking update. The setting in BOINC is the one where the server detects if two connections are within X seconds, then it disables the second and says 'Go away for another X seconds'.

Now I'm off to figure why my i7 machine is only getting 1 WU a time =P

---

The setting in BOINC is the one where the server detects if two connections are within X seconds, then it disables the second and says 'Go away for another X seconds'.

Pretty sure that's the setting I set to 1 minute.

---

[quote]
I'm working on it. Hopefully WUs will be availalble at milkyway_gpu@home sometime this week.

I hope the 6 WUs per cpu will go away at milkyway_gpu@home
At the moment an i7 with a 4830 does more work than a Core 2 Duo with a 4870 - speed does not count but the bigger cache of WUs the i7 receive makes a big difference

---

A BLAST FROM YOUR PAST

The setting in BOINC is the one where the server detects if two connections are within X seconds, then it disables the second and says 'Go away for another X seconds'.

Pretty sure that's the setting I set to 1 minute.

Here's a message that Rosetta (at least used to) send out:

1) 7/3/2006 11:26:15 AM|rosetta@home|Scheduler request to http://boinc.bakerlab.org/rosetta_cgi/cgi succeeded
2) 7/3/2006 11:26:15 AM|rosetta@home|Message from server: Not sending work - last request too recent: 20 sec
3) 7/3/2006 11:26:16 AM|rosetta@home|Deferring communication with project for 4 minutes and 1 seconds

Ideally the setting change should invoke line 2), though as it stands, as far as I can tell, it changes 3) to default to one minute before the exponential back off occurs.

At least on my boxes, I haven't received any equivalent of the line 2) message. Just a heads up Travis, not trying to nag =D

---

I saw this kind of Stop-Message on MW:
- last ... too recent
- wait ... 1 min

This type of message is ok as long as MW dont send any WU of the "ps_sgr_208_test_..." type - even a slow 4830 could calc 6WUs in less then 15sec !

The change does appear to be having a small effect for me.
My rac, which has been hovering just under 40,000 for a while now, jumped by 1100 between yesterday and today.
But my scripts are set to check at 2min intervals and so shouldn't hit your limit anyway.

---

Cheers,

PeterV

.

[quote
1) 7/3/2006 11:26:15 AM|rosetta@home|Scheduler request to http://boinc.bakerlab.org/rosetta_cgi/cgi succeeded
2) 7/3/2006 11:26:15 AM|rosetta@home|Message from server: Not sending work - last request too recent: 20 sec
3) 7/3/2006 11:26:16 AM|rosetta@home|Deferring communication with project for 4 minutes and 1 seconds

Ideally the setting change should invoke line 2), though as it stands, as far as I can tell, it changes 3) to default to one minute before the exponential back off occurs.

At least on my boxes, I haven't received any equivalent of the line 2) message. Just a heads up Travis, not trying to nag =D[/quote]


Like this one ???

6/2/2009 06:53:37|Milkyway@home|Message from server: Not sending work - last request too recent: 22 sec


Seems like it is working just fine.

---

.

Unfortunately the 'Deferring Communication for 1min' isn't a forceful measure - it can still be overridden by clicking update. The setting in BOINC is the one where the server detects if two connections are within X seconds, then it disables the second and says 'Go away for another X seconds'.

Now I'm off to figure why my i7 machine is only getting 1 WU a time =P

6/2/2009 06:57:50|Milkyway@home|Message from server: Not sending work - last request too recent: 33 sec

---

.

Aha ok it's working then :) It just wasn't on my machine. I'll have another peek when I get home tonight. :P

---

Well, for the first time ever I got this message:

02-06-2009 22:39:32 Milkyway@home Requesting new tasks
02-06-2009 22:39:37 Milkyway@home Scheduler request completed: got 0 new tasks
02-06-2009 22:39:37 Milkyway@home Message from server: No work sent
02-06-2009 22:39:37 Milkyway@home Message from server: (reached per-CPU limit of 6 tasks)

So something seems to be working good :)

I like the update, don't change it. The reason my numbers are up is probably because the guys hitting under a minute have been backed off, good move. Patrick

The most work we peanuts have had in a long while.