Collecting custom metrics from redis and sending to Azure Log Analytics

NOTE: Considering that you already have Azure Log Analytics workspace setup and OMS agent running on your server, if you don't:

• Azure Monitor: https://azure.microsoft.com/en-us/services/monitor/
• OMS Agent for Linux: https://github.com/microsoft/OMS-Agent-for-Linux

I decided to use JSON data as it is already supported by OMS Agent, so all you have to do is to configure OMS Fluentd plugin as described here:

• Collecting custom JSON data sources with the Log Analytics agent for Linux in Azure Monitor: https://docs.microsoft.com/en-us/azure/azure-monitor/platform/data-sources-json

The only difference is that your command becomes:

command '/usr/local/bin/redis_info.py'

NOTE: tag oms.api.redismemory

Please refer to https://github.com/flaviotorres/azure-log-analytics-json-redis/blob/master/redis_info.py file and make sure Linux owner and group is set to omsagent:omiusers

Once you have the script in place and running you should see get something like:

{"used_memory": 53858016, "used_memory_rss": 67555328, "keyspace_hits": 3971437, "keyspace_misses": 85502, "expired_keys": 38126, "evicted_keys": 0, "keys": 734}

NOTE: remind to check the log file (/var/opt/microsoft/omsagent/WORKSPACE_ID/log/omsagent.log) after restarting the agent.

Alright, now you should see in Log Analytics Custom Logs a table named redismemory_CL.

Now, here are a few queries for basic monitoring.

// redis keys
redismemory_CL
| project TimeGenerated, keys_d
| render timechart

// redis used memory
redismemory_CL
| project TimeGenerated, used_memory_d, used_memory_rss_d
| render timechart 

// redis keyspace
redismemory_CL
| project TimeGenerated, keyspace_misses_d, keyspace_hits_d
| render timechart 

// Redis Cache hit Ratio
redismemory_CL
| project TimeGenerated, keyspace_misses_d, keyspace_hits_d
| extend ratio = keyspace_hits_d / (keyspace_hits_d + keyspace_misses_d) * 100
| project TimeGenerated, ratio  | render timechart 

Run from bash:

# Adding some random keys
for i in $(seq 10000);do  redis-cli -c -h IP -p PORT  -a PWD set mykey1$RANDOM$i 'helloskdjflaskdjflaajsdfjhkasjdfhaksjdhfakjshdfkajhskfdjhaksjdfhaksjdhfkajshdfkjashdkjfhaskjdhfaksjdhf' EX 60;done

# forcing redis to miss 
for i in $(seq 10000);do  redis-cli -c -h IP -p PORT  -a PWD get mykey11111$i;done

Result: Azure_Log_Analytics_onprem_Redis_Dashboard.png