Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GAP can't fork on systems without overcommit #3509

Closed
ChrisJefferson opened this issue Jun 18, 2019 · 0 comments · Fixed by #3565
Closed

GAP can't fork on systems without overcommit #3509

ChrisJefferson opened this issue Jun 18, 2019 · 0 comments · Fixed by #3565
Labels
kind: enhancement Label for issues suggesting enhancements; and for pull requests implementing enhancements topic: kernel

Comments

@ChrisJefferson
Copy link
Contributor

For the purposes of this issue, there are two important memory values in GAP:

  • SyAllocPool, set with '-s', sets how much memory space GAP asks the OS for.
  • SyStorMax, set with '-o', sets at what amount of memory we warn about excess memory usage and print a warning.

At the moment GAP always sets SyAllocPool is >= SyStorMax.

This gives a problem since we changed SyStorMax to be 3/4 of physical memory, which sets SyAllocPool to 3/4 of physical memory.

This doesn't mean we actually use 3/4 of physical memory, just we tell the OS we might use it. Then when we fork, we are technically using 1.5x physical memory, except again we aren't because (a) most of it is identical and the OS doesn't duplicate identical pages, and (b) most of it is unused.

HOWEVER, you an enable an option which ensures the OS never promises memory it can't give out later, and if you enable that GAP can't fork, which means it can't run external processes.

Travis seems to have some kind of limit against excessive overcommit, from what we can tell, which is causing some travis tests to fail.

My tempted fix to this is to stop making SyAllocPool be >= SyStorMax, and set it back by default to 2g, which was it's default in previous versions. This should, I believe, not make any significant difference to users.

However, I thought I'd write down the issue and see if anyone had any comments.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
kind: enhancement Label for issues suggesting enhancements; and for pull requests implementing enhancements topic: kernel
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants