Hello,

We are running boinc clients on a cluster. We are using boinc to fill up unused time on the cluster, and are trying to adjust the number of cores used dynamically according to the number of our jobs running on a particular node.

The way I am attempting to do this is by setting up a cron job that rewrites the global_prefs_override.xml file every minute according to the number of cpus I'd like boinc to be using, followed by "boinccmd --read_global_prefs_override". Unfortunately, it appears to be ignoring the max_cpus and max_ncpus_pct tag. Other tags work; I can set the max disk space and it is adjusted when the cron job runs. Changing the two cpu tags appears to do nothing, however. Here's my global_prefs_override.xml file:

<global_preferences>
   <run_on_batteries>1</run_on_batteries>
   <run_if_user_active>1</run_if_user_active>
   <start_hour>0.000000</start_hour>
   <end_hour>0.000000</end_hour>
   <net_start_hour>0.000000</net_start_hour>
   <net_end_hour>0.000000</net_end_hour>
   <leave_apps_in_memory>0</leave_apps_in_memory>
   <confirm_before_connecting>1</confirm_before_connecting>
   <hangup_if_dialed>0</hangup_if_dialed>
   <dont_verify_images>0</dont_verify_images>
   <work_buf_min_days>0.100000</work_buf_min_days>
   <work_buf_additional_days>0.250000</work_buf_additional_days>
   <max_cpus>0</max_cpus>
   <cpu_scheduling_period_minutes>60.000000</cpu_scheduling_period_minutes>
   <disk_interval>60.000000</disk_interval>
   <disk_max_used_gb>10.000000</disk_max_used_gb>
   <disk_max_used_pct>50.000000</disk_max_used_pct>
   <disk_min_free_gb>0.100000</disk_min_free_gb>
   <vm_max_used_pct>75.000000</vm_max_used_pct>
   <ram_max_used_busy_pct>50.000000</ram_max_used_busy_pct>
   <ram_max_used_idle_pct>90.000000</ram_max_used_idle_pct>
   <max_bytes_sec_up>0.000000</max_bytes_sec_up>
   <max_bytes_sec_down>0.000000</max_bytes_sec_down>
   <cpu_usage_limit>100.000000</cpu_usage_limit>
</global_preferences>

While I only show max_cpus here, I've tried both tags with the same results.

Is there something else I need to look into here? I cant run the manager because we dont have xwindows set up on the nodes.

This might be helpful too:

19-Nov-2009 13:10:45 [---] Starting BOINC client version 6.6.36 for x86_64-pc-linux-gnu
19-Nov-2009 13:10:45 [---] log flags: task, file_xfer, sched_ops
19-Nov-2009 13:10:45 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8g zlib/1.2.3 c-ares/1.5.1
19-Nov-2009 13:10:45 [---] Data directory: /state/partition1/boinc
19-Nov-2009 13:10:45 [---] Processor: 2 AuthenticAMD AMD Opteron(tm) Processor 246 [Family 15 Model 5 Stepping 10]
19-Nov-2009 13:10:45 [---] Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow
19-Nov-2009 13:10:45 [---] OS: Linux: 2.6.18-92.1.13.el5
19-Nov-2009 13:10:45 [---] Memory: 1.96 GB physical, 996.21 MB virtual
19-Nov-2009 13:10:45 [---] Disk: 103.31 GB total, 98.02 GB free
19-Nov-2009 13:10:45 [---] Local time is UTC -6 hours
19-Nov-2009 13:10:45 [---] Can't load library libcudart
19-Nov-2009 13:10:45 [---] No coprocessors
19-Nov-2009 13:10:45 [---] Not using a proxy

Thanks for any help,
Rye

max_cpus is used by pre-rev-6 BOINC.

IIRC, those tags, which set number of CPUs or GPUs, need a restart of the BOINC client to take effect.

Gruß,
Gundolf

I think zero is read as, revert to global_prefs.xml values.

That was precisely the problem. Thanks a ton.

I think zero is read as, revert to global_prefs.xml values.

Setting options to zero in cc_config.xml means don't use them.

Setting <max_cpus>0</max_cpus> in global_prefs_override.xml on a BOINC 6 system will use the values in global_prefs.xml, since BOINC 6 does not use the <max_cpus> option. It uses the <max_ncpus_pct> option.

The options set in global_prefs_override.xml will only override the same settings as in global_prefs.xml .. when they're not set in the override file, they will always use those values set in the global preferences file.

---