The IQ process is killed by the Linux oom_killer (out of memory killer). This server is dedicated for IQ, therefore it is normal for the IQ service to utilise the CPU resources to the fullest when needed. Presently IQ server is having 64GB memory & 6 CPU.
The currently configured memory for the server is 18 GB for caching, however data loading when a huge transaction is being applied temporarily increases the amount of memory allocated depending on the transaction size to the maximum of shmmax.
Limiting the maximum memory that can be allocated to IQ during these instances by lowering the shmmax value, however this will still not prevent oom_killer from targeting the IQ server, as it will still have a large amount of memory allocated.
How can we configuring the oom_killer to ignore or at least deprioritize the IQ server?