Symptoms
APS 2.0 tasks are failing:
Network error: SSL connection unexpectedly closed. Application endpoint: 192.0.2.2. Request: DELETE ...
where 192.0.2.2 is the application instance endpoint node.
The /var/log/httpd/error_log
file contains Segmentation fault error messages corresponding to task failures by time.
Cause
httpd service is unstable and faces segmentation faults during processing APS 2.0 REST requests.
Resolution
As a workaround, restart Apache service to stop the errors:
# systemctl restart httpd
To allow future root cause diagnostics:
change core pattern:
# mkdir /tmp/coredumps # sysctl -w kernel.core_pattern="/tmp/coredumps/core-%e-sig%s-user%u-group%g-pid%p-time%t" # echo 'kernel.core_pattern="/tmp/coredumps/core-%e-sig%s-user%u-group%g-pid%p-time%t"' >> /etc/sysctl.conf
create
/etc/security/limits.d/core.conf
file with some reasonable limit, e.g. 10 core dumps:* hard core 10 * soft core 10
modify
/etc/systemd/system.conf
and add:DefaultLimitCORE=infinity
modify
/etc/systemd/system/multi-user.target.wants/httpd.service
and add:LimitCORE=infinity
commit the changes:
# systemctl daemon-reexec # systemctl stop httpd # systemctl start httpd