I have the space so I've doubled that to 2.0GB now.

I attempted to install that config file that I copied and pasted from the link but for some reason BOINC manager was then unable to communicate with the client so I removed it again.

Been running HPF successfully all night now without any freeze ups.

I have 2GB of memory in these machines so what would you recommend as a setting for memory usage?

The default for RAM are fine, except running any combination of DDDT/FA@H/HCC would potentially have caused to exceed 1gb disk space (1,000,000,000 byte), particular with HCC having the incorrect 1gb minimum in the work unit header.

The above should have caused messages, but if the flags are off, it would not.

First time i hear cc_config.xml causing a problem.... any wrong line(s) would cause a error message in the log. Create it with an ASCII editor like Notepad making sure it has the exact file name when saved to the BOINC program dir. The Advanced Menu in 5.10.x allowes for direct reading/activation without client restart.

<cc_config>
<log_flags>
<task>1</task>
<file_xfer>1</file_xfer>
<sched_ops>1</sched_ops>
</log_flags>
<options>
</options>
</cc_config>

HPF2 has lower disk needs so that probably explains your success.

Thanking you for posting the settings piece of the log, potentially allowing WCG to put the finger on a problem seen in more places with inexplicable freezes, not only on HPF2 (the looping issue).

I've always had it at 10GB for disk space, a legacy thinking from when on the UD agent, disk space gave a few extra points. On concurrent machines, with 16 CPU cores i think the profile default number should be way way higher for the off-chance.

But, as yet do not know where 1gb came from.... possibly a bug in BOINC picking up a value from the web profile. There have been issues with reading and storing values in 5.10.... i know because my systems use the comma as decimal resulting in outlandish values to be stored.

Hi Sek - good to be back

I too set mine to 10GB of disk space.

Well, i just identified a log problem on my own machine where nothing at all is recorded in 5.10.28 yet BOINCview reporting a job seemingly aborted *reached state "RESULT_ABORTED"

Project Name: FightAIDS@Home
Created: 11/07/2007 02:03:46
Name: faah2636_ZINC01686467_xmd04420_00
Minimum Quorum: 2
Initial Replication: 2

Result Name Status Sent Time Time Due /
Return Time CPU Time (hours) Claimed/ Granted BOINC Credit
faah2636_ ZINC01686467_ xmd04420_ 00_ 3-- Valid 11/10/2007 17:41:35 11/11/2007 12:53:41 11.69 99.1 / 99.1
faah2636_ ZINC01686467_ xmd04420_ 00_ 2-- Valid 11/08/2007 22:26:45 11/10/2007 22:45:49 7.21 67.5 / 99.1
faah2636_ ZINC01686467_ xmd04420_ 00_ 1-- Error 11/08/2007 11:00:00 11/11/2007 14:20:45 0.00 0.0 / 0.0
faah2636_ ZINC01686467_ xmd04420_ 00_ 0-- Error 11/08/2007 10:56:33 11/08/2007 22:22:07 9.26 73.0 / 0.0

If it makes sense to you why the top one of the 4 was sent before mine was aborted, but to think this was a server instruction due redundancy. It happened 1 second after the client reported 3 good results RtR's

Added: In Lone_Wolf's log I'm missing for instance the upload speeds, unless some other none standard flag achieves this.

FWIW I left all those settings at default so it set that by itself.

The only thing I customized under the different profiles was what projects they ran as I had some older machines that couldn't do some of the things offered.

All my machines meet the requirements of the current projects even the oldest PIII 600Mhz machine is crunching DDD without issue.

Now that I've changed the settings I'll try DDD again on this machine and see what happens.