T420 overheats and shuts down while running script

4 risposte [Ultimo contenuto]
amenex
Offline
Iscritto: 01/03/2015

After splitting the source file (1) in a join script whose data is in the 8GB file (2),
the 'puter shuts itself down after running about ten scripts in a set of 83 scripts.

The attached DMESG data indicates that it's trying really hard to protect itself, but
when I stop paying attention, I return to find the 'puter completely off.

I think a brief wait or dummy task inserted between each script to let the CPU's
temperature controller bring the operating temperatures back to normal would allow
the overall script with about sixty steps remaining to finish its task.

Each step takes between 10 and 30 minutes to complete, so a pause of five minutes
between steps would probably keep the overall script going overnight.

The following URL suggests a method of doing this:
https://www.cyberciti.biz/tips/linux-unix-pause-command.html
read -t 300 -p "CPU paused to allow it to cool"
which causes the prompt to disappear for five minutes.

I'll report back after putting the change in effect.

George Langford

AllegatoDimensione
PowerProblems-05312021.DMESG_.txt10.37 KB
amenex
Offline
Iscritto: 01/03/2015

During the first five-minute pause I noticed that the clock speed of the CPU dropped below 1GHz;
but when the next step of the script started after the expected five-minute pause, the speed
jumped to around 3GHz.

Watching the System Monitor, I'm seeing that there are occasional pauses of the usual flat-out
RAM/swap usage, during which the CPU temperatures drop and the CPU clock slows, followed by
quick jumps.

I notice also that the "CPU paused [five minutes] to allow it to cool" messages are stacking
up head-to-tail on the command line. Can I add a \n to the statement to pretty it up ?

The CPU seems to be running too fast while processing the multi-step script; I can't find any
way of under-clocking the CPU in the Add/Remove Applications menu, nor under Power Management
in the Control Center ...

amenex
Offline
Iscritto: 01/03/2015

OK; attached is one complete cycle, during which one script finished, another script ran to
completion, and a five-minute pause began.

The good news is that the 'puter has an optimization routine built in to it, because the
CPU usage ran for long periods at around 90% usage and the clock speed was held to about
66% of full speed, with only a couple of periods at 100% usage and greatly lowered clock
speed.

No further tweaks tonight ...

But here's one of those CPU-hungry scripts:
join -1 1 -2 2 <(sort -k 1,1 Splits-GB-PTR-Labels/Test-05292021-B01-bv.txt) <(sort -k 2,2 GrandSummary/GB-IPv4-PTR-Label-Year-Mo-SortU.txt) | awk '(NF==5){print $3,$1,$2,$4,$5}' '-' > GrandSummary/GB-IPv4-PTR-lebaL-Label-Year-Mo-bv.txt ; read -t 300 -p "CPU paused five minutes to allow it to cool"
"Test-05292021-B01-bv.txt" has 596 rows; "GB-IPv4-PTR-Label-Year-Mo-SortU.txt" has 159,152,578
rows & 8.0 GB; and "GB-IPv4-PTR-lebaL-Label-Year-Mo-bv.txt" has 65,742 rows. The script took
seventeen minutes to run, including the five-minute pause.

AllegatoDimensione
PowerProblems-05312021.DMESG_.II_.txt 8.72 KB
Beformed
Offline
Iscritto: 01/12/2017

I had my X220 being very hot for some period of time running a not very cpu intensive script. Re-pasting it helped

amenex
Offline
Iscritto: 01/03/2015

Followup:
Running the scripts from the same drive on which the data partition is located greatly
increases the speed at which the scripts gather data, as well as reducing the load
on the processor, making the five-minute pauses unnecessary.