I have started building a number of dashboards in Splunk to monitor metrics of our Appian 18.1 on premises environment. One of the files I am using is the perflogs/engine_summary.csv. The fields for each line are as follows:
Timestamp,Gateway Name,Gateway ID,Hostname,Size on Disk (MB),Transaction Count,Chain Status,Uptime (s),Response Time (ms),Total Memory Used (MB)
When validating the accuracy of the Size and Memory metrics, I noticed that the value listed for Total Memory Used for an engine is always lower than what top or ps tells me the process has reserved. The actual memory reserved by the process is usually 45-65% higher than what is reported in the log file.
If the number in the log file doesn't represent the current memory usage for an engine, what does it represent? If it's supposed to represent the current usage, why are the numbers so consistently off?
Thanks in Advance!
Discussion posts and replies are publicly visible
Top has several columns for memory, the main two being virtual and reserved. Are you looking at virtual or reserved in top?
Looking at the reserved and ran ps with -o pid,size parameters. The memory usage is close when an engine is initially started, but after a while, it becomes way off.