| Author |
Message |
|---|
|
|
|
Just curious ... what is the average "wait-time" between finishing a task and reporting the task? |
|
|
|
|
|
I see the tasks are being completed and transferred, but no credit; status just says, "communication deferred" ... is this normal? |
|
|
Volunteer tester
Send message
Joined: 9 Apr 02
Posts: 11987
Credit: 17,853,174
RAC: 58,694
|
|
The reporting actually happens immediately, but the result is batch-uploaded at a later point in time, up to 24 hours later at most, if I remember correctly.
When you see the task is completed and "transferred", that merely means the first stage of reporting happened, but it still has to upload the result. If it cannot upload the result due to the servers being busy or otherwise unable to contact the servers, it will display "communication deferred" as it will automatically retry again in the future. |
|
|
|
|
|
I see; thank you for the info! |
|
|
|
|
The reporting actually happens immediately, but the result is batch-uploaded at a later point in time, up to 24 hours later at most, if I remember correctly.
Or when the nearest deadline in the batch is less than 24 hours away. (Same caveat.)
But I’m pretty sure you have uploading and reporting reversed there. The result files, which are comparatively large, go right away to a simple storage server (barring communication problems). The reports, however, which go to the database server and entail a fair amount of back-end housekeeping work, despite being small in size, are collected in batches for greater efficiency. Grouping the reports reduces the communication overhead from handshaking & negotiation, and I believe also lightens the load on the database, getting more done on each occasion the host & user records are accessed.
____________
  |
|
|
Volunteer tester
Send message
Joined: 9 Apr 02
Posts: 11987
Credit: 17,853,174
RAC: 58,694
|
The reporting actually happens immediately, but the result is batch-uploaded at a later point in time, up to 24 hours later at most, if I remember correctly.
Or when the nearest deadline in the batch is less than 24 hours away. (Same caveat.)
But I’m pretty sure you have uploading and reporting reversed there. The result files, which are comparatively large, go right away to a simple storage server (barring communication problems). The reports, however, which go to the database server and entail a fair amount of back-end housekeeping work, despite being small in size, are collected in batches for greater efficiency. Grouping the reports reduces the communication overhead from handshaking & negotiation, and I believe also lightens the load on the database, getting more done on each occasion the host & user records are accessed.
Yes, that's right. I had them reversed. |
|
|
|
|
|
Any idea how long these uploading and reporting processes should take? It takes days for BOINC to upload and report work units I have finished, and then get new tasks. Message in log is always:
Wed Oct 12 13:10:38 2011 | SETI@home | Sending scheduler request: To fetch work.
Wed Oct 12 13:10:38 2011 | SETI@home | Requesting new tasks for CPU
Wed Oct 12 13:11:54 2011 | | Project communication failed: attempting access to reference site
Wed Oct 12 13:11:54 2011 | SETI@home | Scheduler request failed: Couldn't connect to server
Wed Oct 12 13:11:55 2011 | | Internet access OK - project servers may be temporarily down.
This has been a problem for months, across versions of the program (6.12.35, 6.12.38, 6.12.41). I'm running debugging code related to a different problem - do I need to quit that process (and if so, how?), or is there some other solution? Thanks!
____________
|
|
|
|
|
|
No, the debugging flags have nothing to do with that.
You've probably been bitten by the Hurricane Electric flea.
See the sticky threads on the Number crunching forum.
Gruß,
Gundolf |
|
|
|
|
|
Not sure what the Hurricane Electric flea is, and I didn't see anything about sticky threads on the Number Crunching forum (so they must be called something else there) - but did see the info about trying the proxy servers. I tried that, and I still can't get anything to upload.
Wed Oct 12 22:20:56 2011 | SETI@home | Scheduler request failed: HTTP gateway timeout
Turned off the proxy connection and went back to normal and got the above message in the log (after I reset the project and abandoned the WU that could not upload). Am astonished at how much time I have wasted trying to get this thing to work.
____________
|
|
|
