Monitoring using Huge Pages in Cache (Linux)

Question

Question

Mikhail Khomenko · Jan 3, 2018

Hi, all!
As I know, InterSystems recommends the use of Huge Pages. And if count of Huge Pages is enough, we'll see (in cconsole.log) something like this during Cache startup:

12/29/17-14:40:50:360 (3625) 0 Allocated 4630MB shared memory using Huge Pages: 4096MB global buffers, 256MB routine buffers

But if count of Huge Pages is not enough for location of all Globals and Routines caches, Cache won't use Huge Pages.

The question is: does it exist any metrics (API) in Cache (not in OS), that can show us if Cache uses (or not) Huge Pages currently?
Thanks!

Discussion (2)1

Log in or sign up to continue

Murray Oldfield · Jan 3, 2018

Hi, I do not know of any Caché call. I guess you would be better to use OS commands, eg as simple as:

grep Huge /path/mgr/cconsole.log | tail -1

There is also some discussion and some other commands on a similar topic at;

https://community.intersystems.com/post/linux-transparent-hugepages-and-...é

Regards...

2 0

score 0 · Answer 1 · 2018-01-04T04:24:47-05:00

Hi, Murray! Thanks for an answer!

Yes, that article is brilliant. And I was inspired to monitor as much as we can about Cache memory usage from inside Cache, because it will give us the opportunity to have one "monitoring" class (with a lot of API calls) to gather many Cache-connected metrics. Of course, a lot of metrics we should gather from OS as well.
Ok, let HugePages usage to be an OS-metrics.

Best regards!