BOINC asks for work, then more work, then more work

[Ageless]

To top off the queue of work, BOINC asks for work, then more work, then more work. Why doesn't it ask for the initial amount of work and get it?

01/05/2007 23:02:26|PrimeGrid|Sending scheduler request: To fetch work
01/05/2007 23:02:26|PrimeGrid|Requesting 307 seconds of new work
01/05/2007 23:02:31|PrimeGrid|Scheduler RPC succeeded [server version 508]
01/05/2007 23:02:33|PrimeGrid|File primegrid_llr_wrapper_5.06_windows_intelx86.exe exists already, skipping download
01/05/2007 23:02:33|PrimeGrid|File primegrid_llr_5.06_windows_intelx86.exe.orig exists already, skipping download
01/05/2007 23:02:33|PrimeGrid|[file_xfer] Started download of file llrTPS_2609704
01/05/2007 23:02:33|PrimeGrid|[file_xfer] Started download of file llr.ini.orig
01/05/2007 23:02:34|PrimeGrid|[file_xfer] Finished download of file llrTPS_2609704
01/05/2007 23:02:34|PrimeGrid|[file_xfer] Throughput 975 bytes/sec
01/05/2007 23:02:34|PrimeGrid|[file_xfer] Finished download of file llr.ini.orig
01/05/2007 23:02:34|PrimeGrid|[file_xfer] Throughput 10900 bytes/sec
01/05/2007 23:02:36|PrimeGrid|Sending scheduler request: To fetch work
01/05/2007 23:02:36|PrimeGrid|Requesting 246 seconds of new work
01/05/2007 23:02:41|PrimeGrid|Scheduler RPC succeeded [server version 508]
01/05/2007 23:02:43|PrimeGrid|[file_xfer] Started download of file llrTPS_2609710
01/05/2007 23:02:44|PrimeGrid|[file_xfer] Finished download of file llrTPS_2609710
01/05/2007 23:02:44|PrimeGrid|[file_xfer] Throughput 928 bytes/sec
01/05/2007 23:02:47|PrimeGrid|Sending scheduler request: To fetch work
01/05/2007 23:02:47|PrimeGrid|Requesting 36 seconds of new work
01/05/2007 23:02:52|PrimeGrid|Scheduler RPC succeeded [server version 508]
01/05/2007 23:02:54|PrimeGrid|[file_xfer] Started download of file llrTPS_2609716
01/05/2007 23:02:55|PrimeGrid|[file_xfer] Finished download of file llrTPS_2609716
01/05/2007 23:02:55|PrimeGrid|[file_xfer] Throughput 590 bytes/sec

[davea]

Please set <work_fetch_debug> and send message log.

[mikus]

I reported ticket #151, which was closed as a duplicate of #148.

I have now obtained a message log (edited to remove lines about "skipped", etc.) for a repeat of the situation I posted about, and am re-opening this.

My situation: SIMAP was marked "suspended". I unmarked it and connected. The local client (I'm running a 64-bit Linux version compiled by myself from SVN trunk on JUL 11 2007) issued multiple successively smaller work requests against SIMAP: requesting 57121 seconds, then 23618 seconds, then 11411 seconds, then 3699 seconds, then 1999 seconds (each appeared to result in the scheduling of downloads).

My question is - why not ask for a greater amount of work initially, so as to have fewer follow-on requests ?

[mikus]

message log (edited) with <work_fetch_debug>

[Nicolas]

10/02/2008 16:19:48|PrimeGrid|Sending scheduler request: Requested by user.  Requesting 8844 seconds of work, reporting 2 completed tasks
10/02/2008 16:19:53|PrimeGrid|Scheduler request succeeded: got 13 new tasks
10/02/2008 16:20:08|PrimeGrid|Sending scheduler request: To fetch work.  Requesting 11701 seconds of work, reporting 1 completed tasks
10/02/2008 16:20:13|PrimeGrid|Scheduler request succeeded: got 12 new tasks
10/02/2008 16:20:24|PrimeGrid|Sending scheduler request: To fetch work.  Requesting 6217 seconds of work, reporting 0 completed tasks
10/02/2008 16:20:29|PrimeGrid|Scheduler request succeeded: got 11 new tasks
10/02/2008 16:20:39|PrimeGrid|Sending scheduler request: To fetch work.  Requesting 1113 seconds of work, reporting 0 completed tasks
10/02/2008 16:20:44|PrimeGrid|Scheduler request succeeded: got 3 new tasks
10/02/2008 16:20:55|PrimeGrid|Sending scheduler request: To fetch work.  Requesting 120 seconds of work, reporting 1 completed tasks
10/02/2008 16:21:05|PrimeGrid|Scheduler request succeeded: got 1 new tasks

I'll enable work_fetch_debug and see if I can get this to happen again.

[romw]

Doesn't appear to repro on current builds