climateprediction.net home page
How long before a WU starts sending trickles?

How long before a WU starts sending trickles?

Message boards : Number crunching : How long before a WU starts sending trickles?
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4048 - Posted: 13 Sep 2004, 19:55:27 UTC

I set up a new computer yesterday and added Climate to it. It has been crunching away for a little over a day now and is up to a few % complete on the WU. However, I see now that the WEB site shows that machine has not provided any \"trickles\" back. How long does a WU unit need to run before it starts sending back data?
ID: 4048 · Report as offensive     Reply Quote
ChinookFoehn

Send message
Joined: 7 Aug 04
Posts: 83
Credit: 410,895
RAC: 0
Message 4049 - Posted: 13 Sep 2004, 19:59:50 UTC - in response to Message 4048.  
Last modified: 18 Dec 2004, 8:11:34 UTC

ID: 4049 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 4052 - Posted: 13 Sep 2004, 20:09:01 UTC

A 3.2 GHz Pentium, if that's the one, ought to have trickled several times by now unless it has been preoccupied with other tasks (is it also attached to SETI?)

I suspect a firewall problem - is there anything interesting in the message window about trying to contact the CPDN servers?
ID: 4052 · Report as offensive     Reply Quote
old_user1
Avatar

Send message
Joined: 5 Aug 04
Posts: 907
Credit: 299,864
RAC: 0
Message 4055 - Posted: 13 Sep 2004, 20:13:57 UTC - in response to Message 4052.  

I checked on the servers, no trickles even in the "holding tank" for this host #.
ID: 4055 · Report as offensive     Reply Quote
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4064 - Posted: 13 Sep 2004, 23:03:46 UTC - in response to Message 4052.  
Last modified: 13 Sep 2004, 23:04:52 UTC

> A 3.2 GHz Pentium, if that's the one, ought to have trickled several times by
> now unless it has been preoccupied with other tasks (is it also attached to
> SETI?)
>
> I suspect a firewall problem - is there anything interesting in the message
> window about trying to contact the CPDN servers?
>

Yes, it is the 3.2 GHz system. I am also attached to SETI and it has been "swapping" back and forth between those work units and the climate unit. I will check on how many timesteps it has completed next time it swaps back to CP. So far, BOINC shows that work unit at 1.3% complete, so maybe it just hasn't done enough of the WU yet?

No mssages from my firewall (Zonealarm).... I have BOINC set up as a trusted application. Should I open up any particular ports for CPDN?
ID: 4064 · Report as offensive     Reply Quote
ChinookFoehn

Send message
Joined: 7 Aug 04
Posts: 83
Credit: 410,895
RAC: 0
Message 4065 - Posted: 13 Sep 2004, 23:15:23 UTC - in response to Message 4064.  
Last modified: 18 Dec 2004, 8:12:02 UTC

ID: 4065 · Report as offensive     Reply Quote
old_user845
Avatar

Send message
Joined: 14 Aug 04
Posts: 13
Credit: 1,231,931
RAC: 0
Message 4066 - Posted: 13 Sep 2004, 23:34:23 UTC - in response to Message 4064.  

> So far, BOINC shows that work unit at 1.3% complete, so maybe it just
> hasn't done enough of the WU yet?

Actually your machine will trickle 72 times until your model is finished (24 times in each of the 3 phases). If you divide 100% by 72 trickles, the result is that the first trickle should be reported when 1.389% of the WU are done...
ID: 4066 · Report as offensive     Reply Quote
old_user355

Send message
Joined: 7 Aug 04
Posts: 187
Credit: 44,163
RAC: 0
Message 4067 - Posted: 13 Sep 2004, 23:43:10 UTC - in response to Message 4064.  

> No mssages from my firewall (Zonealarm).... I have BOINC set up as a trusted
> application. Should I open up any particular ports for CPDN?

Nope. If you've already given BOINC permission to access the internet in ZA, that's all you need.

Probably a dumb question, but sometimes you have to ask them. ;) You don't by chance have network access disabled do you? If you do, your trickles are piling up and will be sent when you turn it back on.

<a href="http://www.boinc.dk/index.php?page=user_statistics&amp;project=cpdn&amp;userid=355"><img border="0" height="80" src="http://355.cpdn.sig.boinc.dk?188"></a>
ID: 4067 · Report as offensive     Reply Quote
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4068 - Posted: 13 Sep 2004, 23:47:59 UTC - in response to Message 4067.  

&gt; &gt; No mssages from my firewall (Zonealarm).... I have BOINC set up as a
&gt; trusted
&gt; &gt; application. Should I open up any particular ports for CPDN?
&gt;
&gt; Nope. If you've already given BOINC permission to access the internet in ZA,
&gt; that's all you need.
&gt;
&gt; Probably a dumb question, but sometimes you have to ask them. ;) You don't by
&gt; chance have network access disabled do you? If you do, your trickles are
&gt; piling up and will be sent when you turn it back on.
&gt;
&gt;
No, it's enabled. I've been merrily reporting SETI results back and getting new work units.
ID: 4068 · Report as offensive     Reply Quote
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4101 - Posted: 14 Sep 2004, 12:55:20 UTC

This is getting more curious. Yesterday, BOINC reported that the CPDN WU was at 1.3% complete. This morning, after about 2 hours more CPU time, it now shows at 1.1%. Does it start the WU over every time SETI preempts it?
ID: 4101 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 4104 - Posted: 14 Sep 2004, 14:16:55 UTC - in response to Message 4101.  

&gt; This is getting more curious. Yesterday, BOINC reported that the CPDN WU was
&gt; at 1.3% complete. This morning, after about 2 hours more CPU time, it now
&gt; shows at 1.1%. Does it start the WU over every time SETI preempts it?

No, not unless the model crashed for some reason. Preempting will clear the model from memory by default, so you will revert to the last save point (every 144 timesteps or three model days), but that will only mean about 4 minutes lost work on your machine. You can alter this behaviour in preferences by instructing it to suspend to memory.

What may have happened is that the program detected a possible processing error and so reverted to an earlier stage to rerun that portion of the model.

It also sounds as though most of the time is being spent processing SETI. That may be what you want, but if not you are being affected by a bug in the worksharing element of the BOINC client. You could try altering the preferences or just wait for the next BOINC release.
ID: 4104 · Report as offensive     Reply Quote
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4125 - Posted: 14 Sep 2004, 23:08:51 UTC

Thanks to everyone for their responses. The new machine sent it's first trickle in today, so everything looks okay now.
ID: 4125 · Report as offensive     Reply Quote
old_user1540

Send message
Joined: 26 Aug 04
Posts: 16
Credit: 409,831
RAC: 0
Message 4573 - Posted: 23 Sep 2004, 23:20:25 UTC - in response to Message 4125.  

&gt; Thanks to everyone for their responses. The new machine sent it's first
&gt; trickle in today, so everything looks okay now.
&gt;
&gt;nnnn
<IMG src="http://seti.mundayweb.com/stats.php?userID=423&amp;trans=off"> <img src="http://boinc.mundayweb.com/seti2/stats.php?userID=1424&amp;trans=off">
<img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=333&amp;trans=off"> <img>
ID: 4573 · Report as offensive     Reply Quote

Message boards : Number crunching : How long before a WU starts sending trickles?

©2024 cpdn.org