Extend MongoDB Memory Limit

This procedure requires downtime of the application stack

All values in Red are variables and are subject to change as per requirement

Symptoms

The default deployment comes with a standard workload in mind and as the size of dataset grows beyond the limit of standard deployment profile, the mongoDB response becomes noticeably slow.  Some of the symptoms are; slow response in data retrieval from historical data, or the currently ongoing operations become very slow. 

Some of the indications that MongoDB  performance is suffering  because of low memory, can be actually seen in logs as well. For Example

serverStatus was very slow: 

OR 

Out of memory: Kill process 13130 (mongod) 
server4 kernel: [1731430.441717] Killed process 13130 (mongod) 

Its time to perform a memory upgrade for MongoDB Service.

MongoDB Memory Reservation  Algorithm

MondoDB's uses a thumb rule for using memory

1. (50% of RAM) - 1 GB
2. 256MB

and uses which ever is higher. This is valid for any environment, whether running inside a container or as a standard Host Service, until customised.

Procedure

Calculate the MongoDB current Dataset by using below given procedure.

perform this procedure on the HOST system

# docker exec -it expertflow_mongo_1 sh


Copy and paster the code between "----CODE BLOCK----" lines on the mongo terminal

----CODE BLOCK----
mongo --eval "db.serverStatus();"|\
awk '/maximum bytes configured"/ { MBC=$NF } 
/"bytes currently in the cache"/ { BCITC=$NF} 
END {if(BCITC > MBC){
printf "Memory Upgrade required  %s MB \n", BCITC/1024
} else {
print "no upgrade needed"
}
} '
----CODE BLOCK----

if this above given procedure says "Memory Upgrade required", please note the recommendation figure.

please configure the MongoDB's memory utilisation as per below.

Procedure

Simplified Deployment

• edit the  DEPLOYMENT_PATH/sds/docker-compose-core.yml  for single node deployment or DEPLOYMENT_PATH/sds/docker-compose-ha-mongo.yml in case HA deployment
  
  • change these parameters to suitable value.this value should be in multiples of 256 like 512,768,1024 and so on. 

    • mem_limit: (X)m   – replace X with value

    • memswap_limit: (X*2)m – replace X with value

• After changing the values appropriate to the environment, please restart or just up -d the mongoDB service container.
  • DEPLOYMENT_PATH/eftasker service, select corresponding number of service mongo and then select "up"

Manual Deployment

• edit the  chat-solution/docker/docker-compose-core.yml and locate the service of mongo
  • change these parameters to suitable value.this value should be in multiples of 256 like 512,768,1024 and so on. 

    • mem_limit: (X)m   – replace X with value

    • memswap_limit: (X*2)m – replace X with value

• restart the solution 
  • docker-compose -f chat-solution/docker/docker-compose-core.yml down
  • docker-compose -f chat-solution/docker/docker-compose-core.yml up -d