Hello tfmagnetism, Is this a laptop? Do you have Leave Application In Memory selected? Is the CPU set to run BOINC 100% or does it turn BOINC projects off every few seconds?

Just trying to figure out what might be causing strange behavior.
Lawrence

Hi Lawrence,

No it's a desktop. LAIM is on. Not quite sure what you mean about the cpu, but it's all on 100%, with a period of 0.04 minutes idle time. Ah you guys have been holding out on me lol! From what I just read in the forums this is all normal (ish)! A checkpoint only at 11-12mins cpu time, and job 2 must be taking an eternity until it makes another checkpoint. What's WEIRD is why this only happens in about 50% of WUs. I've been reading that it's normal for job 2 to take hours before checkpointing again. Anyway heres my global prefs if you wan't to have a look:

<global_preferences>
<source_project>http://www.worldcommunitygrid.org/</source_project>
<source_scheduler>https://grid.worldcommunitygrid.org/boinc/wcg_cgi/fcgi</source_scheduler>
<mod_time>1328483436</mod_time>
<cpu_scheduling_period_minutes>120</cpu_scheduling_period_minutes>
<disk_interval>60.0</disk_interval>
<disk_max_used_gb>10.0</disk_max_used_gb>
<disk_max_used_pct>50.0</disk_max_used_pct>
<disk_min_free_gb>0.5</disk_min_free_gb>
<end_hour>0</end_hour>
<idle_time_to_run>0.04</idle_time_to_run>
<leave_apps_in_memory/>
<max_bytes_sec_down>0.0</max_bytes_sec_down>
<max_bytes_sec_up>0.0</max_bytes_sec_up>
<daily_xfer_period_days>0</daily_xfer_period_days>
<daily_xfer_limit_mb>0.0</daily_xfer_limit_mb>
<max_cpus>16</max_cpus>
<max_ncpus_pct>100.0</max_ncpus_pct>
<suspend_cpu_usage>0.0</suspend_cpu_usage>
<net_end_hour>0</net_end_hour>
<net_start_hour>0</net_start_hour>
<start_hour>0</start_hour>
<cpu_usage_limit>100.0</cpu_usage_limit>
<ram_max_used_busy_pct>50.0</ram_max_used_busy_pct>
<ram_max_used_idle_pct>75.0</ram_max_used_idle_pct>
<vm_max_used_pct>75.0</vm_max_used_pct>
<work_buf_min_days>0.0</work_buf_min_days>
<work_buf_additional_days>0.0</work_buf_additional_days>
<suspend_if_no_recent_input>20.0</suspend_if_no_recent_input>
</global_preferences>

Sekerob,
Not sure what that's going to achieve since this seems to be a checkpoint problem like what several others have been reporting. But I'll certainly consider it.

I'm going to look into hibernation OR boosting CEP2 a bit before shutdown until it makes a checkpoint.

In BOINC, slots, 0 folder I get in stdout:

[00:18:31] [INFO] Skipping checkpoint per user settings.
[00:21:05] [INFO] Checkpoint complete.
[00:32:16] [INFO] Checkpoint complete.

But I'm sure that's nothing to worry about. A bit worried about the "skipping" bit there.

This from client_state in BOINC directory:

<active_task>
<project_master_url>http://www.worldcommunitygrid.org/</project_master_url>
<result_name>E206605_695_C.24.C21H16N2Si.01364647.4.set1d06_0</result_name>
<active_task_state>1</active_task_state>
<app_version_num>640</app_version_num>
<slot>0</slot>
<checkpoint_cpu_time>735.310714</checkpoint_cpu_time>
<checkpoint_elapsed_time>812.553997</checkpoint_elapsed_time>
<checkpoint_fraction_done>0.000000</checkpoint_fraction_done>
<checkpoint_fraction_done_elapsed_time>0.000000</checkpoint_fraction_done_elapsed_time>
<current_cpu_time>3578.678500</current_cpu_time>
<once_ran_edf>0</once_ran_edf>
<swap_size>226050048.000000</swap_size>
<working_set_size>67145728.000000</working_set_size>
<working_set_size_smoothed>67145728.000000</working_set_size_smoothed>
<page_fault_rate>14.694281</page_fault_rate>
</active_task>


Now that above was an updated client_state when the other WU running today, non CEP2, has just downloaded. So quite why it shows 0% fraction done - seems a strange but perhaps that's normal.

This CEP2 is certainly a strange one I'll say that!

Hi Lawrence,

I just pushed the boat out, and ran up to 30% and it just "checkpointed". So it is working after all, just the checkpoint of job 2 (or something) must be really variable! Wow, my other 50% WUs were checkpointing so that I didn't have problems shutting down/restarting even after only several percent. CPU time 3h:34m at checkpoint, elapsed time 3h:45m! Wow. Oh my god that's quite a jump from a few percent. That's a really variable checkpointing system there.
boinc_task_state ( BOINC,slots,0 folder) gives:

<active_task>
<project_master_url>http://www.worldcommunitygrid.org/</project_master_url>
<result_name>E206605_695_C.24.C21H16N2Si.01364647.4.set1d06_0</result_name>
<checkpoint_cpu_time>12890.737033</checkpoint_cpu_time>
<checkpoint_elapsed_time>13330.422573</checkpoint_elapsed_time>
<fraction_done>0.298397</fraction_done>
</active_task>

Now I'm understanding things. I better undelete that hibernation file that I got rid of - I never use hibernation. Wow that's lots of lost work for a lot of people presumably. Now we're really understanding things better. But it only seems to be so noticeable for around 50% of WUs. I think I'll have another look at the FAQs later and see if that's made clear. You see, we only do about 20-25% usually of CEP2 before we shutdown again. OK well at least now I know what's happening.