Tim,
when I tried to find info about hugepage, oracle-base.com poped up on the top list. I am so glad to know you are export on hugepage.
Our env. is 11gr2 RAC, 4 nodes and we use hugepage to help the 10tb import process.
1. we have 135gb RAM on OS which is RHEL
2. instance SGA_max_size is 70gb, other sga and mem params are set to enable hugepage
3 we have HugePages_Total: 34305, HugePages_Free: 705 HugePages_Rsvd: 705 Hugepagesize: 2048 kB, this come up to 68gb hugepage size
4. summarize all above, the total size of hugepage = SGA_max_size +7gb
5. OS always have 77 gb RAM in use which is size of hugepage, also I think hugepage ping them in memory, right?.
Here is my question,
1. I always watch the total of ram in use, it never gets over 77gb, no matter I have big load to database or not. sga_max_size takes 70gb in this RAM, so it only have 7gb for other processes. sometimes I saw database wait time is 99%, it mean database is hung, why can't use more RAM on OS, that is not fair, right?
2. I always puzzled by this 135 gb RAM on OS, why only 77gb RAM in use, what does stop using more RAM when needed?
3. I thought to increase the size of kernel.shmmax, but you told me not to do so. I agreed on this param, but which param can be used to max RAM usage?
thank you so much.
RB