Details
-
Improvement
-
Resolution: Done
-
Major - P3
-
None
-
All
Description
The description of page fault as observed in mongostat and in serverStatus is not 100% clear and often leads O/S oriented people to misinterpret the meaning. The docs glossary def inition of page fault is even less appropriate for the mongoDB meaning. The description everywhere should always compare and contrast our page fault def to the more tradition def so that it is very very clear that our page faults do not necessarily imply a yellow flag for poor system performance or configuration. The def should look something like this:
page_fault:
Reports the total number of page faults in one second.
Do not confuse basic operating system / physical memory page
fault behavior with mongoDB page faults. In basic operating system
parlance, page faults occur when physical memory is exhausted and pages of physical memory are swapped to disk. In mongoDB, page faults occur when operations (both read AND write) access parts of the memory mapped files that are not currently available in physical memory. The faults counter may increase dramatically during moments of poor performance and may correlate with limited physical memory environments and accessing much larger data sets (like scanning an entire collection). Limited and sporadic page faults do not
necessarily indicate a problem or a need for tuning.
Attachments
Issue Links
- is duplicated by
-
DOCS-3346 Comment on: "manual/administration/monitoring.txt"
-
- Closed
-