Server Slient Crash – ScadaServer killed by OOM killer on Linux

[JW]

ScadaServer killed by OOM killer on Linux
Meanwhile, the ScadaComm service is still running.

The service is started on 17Apr and killed on 11May
https://1drv.ms/i/s!Akf5dL_rRmTTjjIapTXo9QQovpqB?e=ew5Ogi

Manual restart on 23May. No detail of the crash in scadaserver log file
https://1drv.ms/i/s!Akf5dL_rRmTTjjGBBE6c0YA5LPKa?e=Bjzdur

Memory usage seems normal after restarting the service in servral hours
https://1drv.ms/i/s!Akf5dL_rRmTTjjOoYEuXvna9pwDg?e=iHCjYr

Have anyone encounter this issue before?

This issue has appeared twice on the same server. Similar projects are running on similar servers, only this one have this issue.

[JW]

As temporary measure, is there any recommended way to monitor and auto restart the service on crash?

For systemd daemon file, it’s easy to add

[Service]
Restart=always
RestartSec=30

But the Scada 5.8.4 is using init.d, and the daemon file is a complicated script. Is there any similar to do it?

[Mikhail]

Hi,

Some thoughts:
– Server v5 requires a lot of RAM when trends (charts) are requested by a user.
– Is it possible to check what amount of RAM was consumed by the Server service when it was killed?
– 218 MB of free memory is not big. If possible, add memory to the server.
– Rapid SCADA 6 uses systemd daemons. Consider moving to v6 in the future.

[Author]

Posts