-
Type: Bug
-
Resolution: Done
-
Priority: Minor - P4
-
Affects Version/s: 3.2.8, 3.3.9
-
Component/s: Testing Infrastructure
-
Fully Compatible
-
ALL
-
TIG 2017-03-06
We run dmesg -T > /dev/null 2>&1 instead of $sudo dmesg -T > /dev/null 2>&1 to see if human readable timestamps are supported.
# Print out any Out of Memory killed process messages. - command: shell.exec params: system_log: true working_dir: src # Temporary files created in src will be cleaned up in "pre". script: | ${set_sudo} # Use dmesg -T option, if supported, to display timestamps. dmesg=dmesg dmesg -T > /dev/null 2>&1 if [ $? -eq 0 ]; then dmesg="dmesg -T" fi $sudo $dmesg 2> /dev/null > dmesg.txt if [ $? -ne 0 ]; then echo "Cannot check for OOM (Out of memory) killed processes on this platform" exit 0 fi egrep -i '(Out of memory|OOM[- ]killer|Killed process)' dmesg.txt > oom.txt if [ -s oom.txt ]; then echo "OOM (Out of memory) killed processes detected" cat oom.txt else echo "No OOM (Out of memory) killed processes detected" fi
- related to
-
SERVER-24211 Addition of oom killed processes to the system logs tab
- Closed