Swap Memory - WARNING

[cyberfinder]

 ran  healthcheck script  in MDS and receiving following error. could you please advice whether its is normal or critical behavior. 

Swap Memory - WARNING Critical: Swap memory is 17% used.

free -g -t
total used free shared buff/cache available
Mem: 125 48 2 1 74 74
Swap: 66 11 54
Total: 191 59 57

[_Val_]

As the message says, "Critical"

[cyberfinder]

Any recommendation  to resolve this issue ?

[Timothy_Hall]

There is no issue.  You have a total of 128GB of RAM which is actively utilizing 48GB for code execution.  74GB more is currently available for code execution.  Buffering/caching is currently consuming that 74GB to optimize hard drive operations, but that memory can be freed at a moment's notice and reallocated for code execution.

What the healthcheck script is flagging is that at some point the system ran low on free memory and allocated 11GB in swap space.  Whatever situation caused the low memory condition has abated, which is why there is now a healthy 74GB available for code execution allocation.  However even after the low memory situation is resolved, any swap space allocated during that period is not relinquished back in anticipation of that situation possibly happening again, and that swap space will remain allocated until the next system boot. 

There is no way to look only at the swap counters and determine when the low memory situation occurred or even if it is occurring right now (so the healthcheck script just reports it with no further context), however if the swap ballooning  happened in the last 30 days you should be able to use the sar -S -f /var/log/sa/saXX command (XX is the day number of the month) to look back and see when the swap space allocation jumped. You could also see this using the historical mode of cpview with the -t option.

[cyberfinder]

Thanks for reply observed swap ballooning, every month it is increasing 2-3 percent. anyhow ,let me try to get outputs of the provided commands. 

[CheckPointerXL]

same behavoir in a lot of customer.

I recently upgraded a lot of cluster to 81.10 to a single customer and they started to be flooded with swap warning; anyway, ram usage is ok in all situation

any workaround to contain swap usage ? it seems that in 81.10 free memomry is working a little bit worser, am i wrong?

[_Val_]

Are we still talking about a management server? Or GWs? In either case, please open a TAC request for swap issues.

[VSX_Bernie]

@CheckPointerXL  - Did you ever come to any conclusion in regards to this?

We experience a steady increase in swap use on all SMS servers (both MDS and Log Server), when they are running servicing logs.

We are also running R81.10 - but this seems to be an issue that has been around on since R80.20 at least:
R80.20 (mlm) log servers swapping - Check Point CheckMates

You can also find my in-depth explanation of our issue in this.

[CheckPointerXL]

the conclusion is that for mgmt/log server swapping is a "normal" behavoir.... just pay attention on GW side about swapping

if swap is out of control on mgmt/log server TAC  case is needed to investigate

[VSX_Bernie]

All right - thanks for sharing