climateprediction.net (CPDN) home page
Thread 'No server responding'

Thread 'No server responding'

Message boards : climateprediction.net Science : No server responding
Message board moderation

To post messages, you must log in.

AuthorMessage
Stan Wells

Send message
Joined: 15 Nov 04
Posts: 6
Credit: 1,057,562
RAC: 0
Message 10130 - Posted: 1 Mar 2005, 4:14:27 UTC

I finished up my last work unit on Thursday 02/24 and have been unable to get a new work unit. the message that comes up says that "no schedulers responded". Being Monday evening, I now wonder if something else is wrong. It seemed to upload the old work unit, but the work tab says hadsm3 4.04 is "ready to report". Is something else wrong?
ID: 10130 · Report as offensive     Reply Quote
ProfileThyme Lawn
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1283
Credit: 15,824,334
RAC: 0
Message 10131 - Posted: 1 Mar 2005, 7:34:01 UTC

Hi Stan,

From your stats pagee it looks like you're tryimg to upload all the trickles from the model at the same time. Unfortunately there's been a problem with uploading lots of trickles in one go for at least 4 weeks.

The symptoms of the problem are:

1) Every CPDN scheduler request gets a <b>No schedulers responded</b> message. This is separate from the BOINC 4.19 problem that sometimes parses the scheduler reply incorrectly.

2) Your computer and results pages <i>may</i> show different dates for the last trickle received (but not in your case). If they do the computer page will show the last trickle the scheduler generated a reply for and the results page will show the last trickle that was processed.

3) You'll have lots of <b>trickle_up_*.xml</b> files (usually over 40, but it may be less) in your <b>projects/climateprediction.net</b> directory.

The work around is to create a directory and move all but 30 of the trickle_up_*.xml files into that directory, do an update (which should succeed and remove the files you left behind), then move the other files back in batches of 30 followed by an update each time.
"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer
ID: 10131 · Report as offensive     Reply Quote
Stan Wells

Send message
Joined: 15 Nov 04
Posts: 6
Credit: 1,057,562
RAC: 0
Message 10139 - Posted: 1 Mar 2005, 14:15:37 UTC - in response to Message 10131.  

Thanks for the response. That did take care of the problem. I am currently not running the 4.19 BOINC - any reason that I should upgrade to that version before my next work unit?

stan

&gt; Hi Stan,
&gt;
&gt; From your stats pagee it looks like you're tryimg to upload all the trickles
&gt; from the model at the same time. Unfortunately there's been a problem with
&gt; uploading lots of trickles in one go for at least 4 weeks.
&gt;
&gt; The symptoms of the problem are:
&gt;
&gt; 1) Every CPDN scheduler request gets a <b>No schedulers responded</b> message.
&gt; This is separate from the BOINC 4.19 problem that sometimes parses the
&gt; scheduler reply incorrectly.
&gt;
&gt; 2) Your computer and results pages <i>may</i> show different dates for the
&gt; last trickle received (but not in your case). If they do the computer page
&gt; will show the last trickle the scheduler generated a reply for and the results
&gt; page will show the last trickle that was processed.
&gt;
&gt; 3) You'll have lots of <b>trickle_up_*.xml</b> files (usually over 40, but it
&gt; may be less) in your <b>projects/climateprediction.net</b> directory.
&gt;
&gt; The work around is to create a directory and move all but 30 of the
&gt; trickle_up_*.xml files into that directory, do an update (which should succeed
&gt; and remove the files you left behind), then move the other files back in
&gt; batches of 30 followed by an update each time.
&gt;
ID: 10139 · Report as offensive     Reply Quote
ProfileThyme Lawn
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1283
Credit: 15,824,334
RAC: 0
Message 10142 - Posted: 1 Mar 2005, 15:43:38 UTC - in response to Message 10139.  

&gt; Thanks for the response. That did take care of the problem. I am currently
&gt; not running the 4.19 BOINC - any reason that I should upgrade to that version
&gt; before my next work unit?

Version 4.19 does fix a couple of problems relating to CPDN but it also introduces the occasional reply parsing error I mentioned (the request is periodically retried, so nothing is actually lost). The fixes are:

1. Improved resilience of file upload and download (particularly relevant to the initial program download, but probably of no great advantage to you as you've already got the current versions on your system).

2. If a BOINC benchmark is due when the program starts up it's now run before attempting to start any applications. Previously you could lose a CPDN model if BOINC tries to suspend the app before it's fully started up. That may be more relevant to you.

There's no requirement to upgrade, but if you decide to do so the best thing is to suspend and exit BOINC, make a backup of your BOINC directory (just in case anything goes wrong), install the new version over the top of the existing one, restart BOINC. If everything's fine delete your backup, otherwise stop BOINC, restore your backup and restart BOINC again.
"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer
ID: 10142 · Report as offensive     Reply Quote

Message boards : climateprediction.net Science : No server responding

©2024 cpdn.org