You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
jkrems opened this issue
Dec 18, 2015
· 2 comments
Labels
duplicateIssues and PRs that are duplicates of other issues or PRs.memoryIssues and PRs related to the memory management or memory footprint.vmIssues and PRs related to the vm subsystem.
There's an interesting possible problem when upgrading from node 0.10 to node 4: It seems like running code via vm.runInNewContext retains the whole context, including all native constructors, globals, etc.. From our current testing it looks like those are never collected even after nothing in our app code holds a reference to any object belonging to the context, leading to the process running out of memory eventually. After we stopped using vm.runInNewContext, the problem went away.
As mentioned above, we have only circumstantial evidence that the leak is in vm.runInNewContext itself.
The text was updated successfully, but these errors were encountered:
ChALkeR
added
duplicate
Issues and PRs that are duplicates of other issues or PRs.
vm
Issues and PRs related to the vm subsystem.
memory
Issues and PRs related to the memory management or memory footprint.
labels
Dec 18, 2015
duplicateIssues and PRs that are duplicates of other issues or PRs.memoryIssues and PRs related to the memory management or memory footprint.vmIssues and PRs related to the vm subsystem.
There's an interesting possible problem when upgrading from node 0.10 to node 4: It seems like running code via
vm.runInNewContext
retains the whole context, including all native constructors, globals, etc.. From our current testing it looks like those are never collected even after nothing in our app code holds a reference to any object belonging to the context, leading to the process running out of memory eventually. After we stopped usingvm.runInNewContext
, the problem went away.As mentioned above, we have only circumstantial evidence that the leak is in
vm.runInNewContext
itself.The text was updated successfully, but these errors were encountered: