Memory FULL in Production

[surya sekhar]

Hi,

Good Morning,

In production we are using two node Cluster. Memory utilization is almost 100% in both the nodes. how to overcome this approch..

Thanks in Advance!

Surya.

[Abhishek Mandhyani]

It really depends on various factors.
1. Latency and number of calls per node.
2. Policy code
3. Logging
4. Active listeners.
Have a look at this document and try to find out the root cause.
https://techdocs.broadcom.com/content/broadcom/techdocs/us/en/ca-enterprise-software/layer7-api-management/api-gateway/9-3/learning-center/thinking-in-policy/policy-performance-optimization.html

Regards
Abhi

[surya sekhar]

Thanks Abhi for the Quick response..

Regards,
Surya

[David Armistead]

Linux will use as much memory as it can for buffers and cache, so it will always appear "full" after a period of time as more things are cached. Buffers/cache can always be reclaimed for use by programs that need more memory. So, from the screenshot you provided, there is actually 19G available for use. No need to worry, all is well. 

-David Armistead
First Citizens Bank
Raleigh, NC

[surya sekhar]

Thanks David For Quick Response.

Regards,
Surya.

[Jay Thorne]

That's a misunderstanding of that data. In Linux systems, all unused RAM is allocated for write buffers and read buffers. 

So your system has 1.0 G of disk IO write buffers, and 18.0 G of Disk IO read buffers. That means that the gateway process is using probably around 13 G RAM, which is perfectly normal. If the gateway process needs more RAM, the OS will shrink the read and write buffers.

------------------------------
Architect
Broadcom
------------------------------

[surya sekhar]

Thanks Jay for Quick response, so everything looks cool, so no need to take any action rite?

Regards,
Surya.

[Jay Thorne]

No, if anything, this instance has too much RAM. 19G free of 32 is a bit over-provisioned.

------------------------------
Architect
Broadcom
------------------------------