Double tap contact

Message boards : Number crunching : Double tap contact

To post messages, you must log in.

AuthorMessage
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 2354 - Posted: 22 Nov 2012, 21:30:06 UTC
Last modified: 22 Nov 2012, 21:30:38 UTC

I noticed that my BOINC 7.0.38 is doing a double contact each time it contacts Enigma on work requests and reporting of work, as the first time it gets an error 403, then it immediately redoes a contact which goes through.

Something like this:
22/11/2012 22:22:06 | Enigma@Home | [sched_op] Starting scheduler request
22/11/2012 22:22:06 | Enigma@Home | Sending scheduler request: To report completed tasks.
22/11/2012 22:22:06 | Enigma@Home | Reporting 1 completed tasks
22/11/2012 22:22:06 | Enigma@Home | Not requesting tasks: project is not highest priority
22/11/2012 22:22:06 | Enigma@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
22/11/2012 22:22:06 | Enigma@Home | [sched_op] ATI work request: 0.00 seconds; 0.00 devices
22/11/2012 22:22:07 | Enigma@Home | Scheduler request failed: Error 403
22/11/2012 22:22:07 | Enigma@Home | Sending scheduler request: To report completed tasks.
22/11/2012 22:22:07 | Enigma@Home | Reporting 1 completed tasks
22/11/2012 22:22:07 | Enigma@Home | Not requesting tasks: project is not highest priority
22/11/2012 22:22:07 | Enigma@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
22/11/2012 22:22:07 | Enigma@Home | [sched_op] ATI work request: 0.00 seconds; 0.00 devices
22/11/2012 22:22:10 | Enigma@Home | Scheduler request completed
22/11/2012 22:22:10 | Enigma@Home | [sched_op] Server version 701
22/11/2012 22:22:10 | Enigma@Home | Project requested delay of 11 seconds
22/11/2012 22:22:10 | Enigma@Home | [sched_op] handle_scheduler_reply(): got ack for task wrmkx-auto-110-o18-d1_0_13731038_25_0
22/11/2012 22:22:10 | Enigma@Home | [sched_op] Deferring communication for 11 sec
22/11/2012 22:22:10 | Enigma@Home | [sched_op] Reason: requested by project

Something to do with the proxy, Slawek?
Jord.

BOINC FAQ Service.
ID: 2354 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,981,757
RAC: 386,200
Message 2364 - Posted: 24 Nov 2012, 16:50:36 UTC - in response to Message 2354.  
Last modified: 25 Nov 2012, 9:37:52 UTC

I think that your client uses very old project config. At some point the project had two schedulers, but since my ISP does wild things with IP addresses/pools I had to remove one.
I believe it was the first entry in schedulers.txt so probably that's why the client tries to contact the wrong URL first and then it falls back to the other one.
M4 Project homepage
M4 Project wiki
ID: 2364 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 2366 - Posted: 24 Nov 2012, 22:42:27 UTC - in response to Message 2364.  
Last modified: 24 Nov 2012, 23:20:03 UTC

Ah, I see. The old one I presume being
<scheduler_url>http://kfd.no-ip.org/enigma_cgi/cgi</scheduler_url>

and the new one
<scheduler_url>http://www.enigmaathome.net/enigma_cgi/cgi</scheduler_url>


Edit: Yep, that did it.
Now though...
24/11/2012 23:43:25 | Enigma@Home | Requesting new tasks for CPU
24/11/2012 23:43:25 | Enigma@Home | [sched_op] CPU work request: 13783.97 seconds; 0.00 devices
24/11/2012 23:43:25 | Enigma@Home | [sched_op] ATI work request: 0.00 seconds; 0.00 devices
24/11/2012 23:43:28 | Enigma@Home | Scheduler request completed: got 0 new tasks
24/11/2012 23:43:28 | Enigma@Home | [sched_op] Server version 701
24/11/2012 23:43:28 | Enigma@Home | Project has no tasks available
24/11/2012 23:43:28 | Enigma@Home | Project requested delay of 11 seconds
24/11/2012 23:43:28 | Enigma@Home | [sched_op] Deferring communication for 11 sec
24/11/2012 23:43:28 | Enigma@Home | [sched_op] Reason: requested by project

;-)

Edit two:
For others with the same problem, it's probably easiest to set NNT and run your cache dry, report all, the remove the project and add it again.
But if you don't mind editing some files, then do the following:
- Exit BOINC completely, make sure it isn't running anymore (boinctray.exe can stay on).
- Navigate to your BOINC Data directory and edit master_www.enigmaathome.net.xml with something like Wordpad (that way the contents show orderly). Press CTRL + F and type 'kdf' (without quotes) in the search bar. Press Enter.

Where it finds the URL http://kfd.no-ip.org/enigma_cgi/cgi, remove the line starting with <!-- and remove the line stating <link rel="boinc_scheduler"
Save changes to the file (CTRL + S) and close this file.

- Open client_state.xml with Wordpad.
Press CTRL + F
Type Enigma in the search bar and press Enter.
Scroll down a bit.
Where it says:
<scheduler_url>http://kfd.no-ip.org/enigma_cgi/cgi</scheduler_url>
<scheduler_url>http://www.enigmaathome.net/enigma_cgi/cgi</scheduler_url>

Remove just the line <scheduler_url>http://kfd.no-ip.org/enigma_cgi/cgi</scheduler_url>, removing the empty line as well. The only scheduler showing should be the enigmaathome one.
Save changes (CTRL + S) and exit this file.

Now restart BOINC. If need be, force a contact with Enigma (Update) to check your changes.
Jord.

BOINC FAQ Service.
ID: 2366 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,981,757
RAC: 386,200
Message 2367 - Posted: 25 Nov 2012, 9:43:33 UTC - in response to Message 2366.  

I'm surprised that the BOINC client didn't correct it by itself.
As far as I know after 10 failed contacts it pulls down fresh schedulers list (which is hidden in project's index.php). It looks like only a complete failure (all schedulers down) triggers the mechanism.


M4 Project homepage
M4 Project wiki
ID: 2367 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 2370 - Posted: 25 Nov 2012, 20:16:00 UTC - in response to Message 2367.  

I think that because it ran two schedulers in a row, the faulty and the good one, that it didn't see that it went wrong all those times, since the second contact was always correct.

Reporting it to the alpha list though. :-)
Jord.

BOINC FAQ Service.
ID: 2370 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,981,757
RAC: 386,200
Message 2372 - Posted: 26 Nov 2012, 9:44:46 UTC - in response to Message 2370.  

IMO it should be corrected. If one of the schedulers regularly fails then the client should verify if it is still on the list. My old scheduler returns 403 so it fails immediately, http timeout would be much worse due to additional delay while waiting.

M4 Project homepage
M4 Project wiki
ID: 2372 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 2373 - Posted: 26 Nov 2012, 12:19:46 UTC - in response to Message 2372.  
Last modified: 26 Nov 2012, 12:20:38 UTC

Hold on, didn't you change the project URL only back in 2007? When did you change the scheduler URL, from kfd.no-ip.org, do you remember? (It's not in the news, I checked).

Could it have been that for a (short) time, there were two scheduler URLs on the front page? As that seems to be the only way that the master file could've gotten these. However, it's been shorter than 5 years since I last ran Enigma... (so recheck the master file sent to use, just to be sure.) ;-)
Jord.

BOINC FAQ Service.
ID: 2373 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,981,757
RAC: 386,200
Message 2374 - Posted: 26 Nov 2012, 12:55:31 UTC - in response to Message 2373.  
Last modified: 26 Nov 2012, 12:58:37 UTC

The project URL changed only once and that was in late 2007.
However, the scheduler URL(s) were changed a few times.
At some point, I think it was late 2010, I set up dual scheduler config. Back then the project had two schedulers that worked at the same time, so yes, there were two entries in schedulers.txt (which is then included in index.php).
I have no idea how the clients decided which one to contact but eventually it was pretty well balanced (near 50% traffic on each of them).
To be honest I don't remember when I removed one of the schedulers, but I'd guess it was early summer 2011.
M4 Project homepage
M4 Project wiki
ID: 2374 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Double tap contact




Copyright © 2017 TJM