Steps to Reproduce:
Does this issue occur when all extensions are disabled?: No, requires the TSLint extension (tested with v1.2.3)
Additional Notes:
Both regular and insiders editions have this issue.
VSCode 1.39.2 appears to be working correctly, I've been unable to reproduce. RAM usage will increase but never shoots up so rapidly and if left alone for a few seconds the GC will kick in and usage drops drastically.
It appears these issues might be related:
There are a number of other issues mentioning VSCode being "slow" in the last few weeks which leads me to believe this issue is affecting quite a few devs.
Saw similar issue in my setup also
Code Version: 1.40.2
OS Version: win 10


Please file this issue against the tslint extension since it seems to cause this
This issue should not be closed. The same version of TSLint works perfectly fine on older versions of VSCode which indicates the extension itself is unlikely the cause.
The change might be related to how VSCode starts a new process. The parameters passed to start the tsserver.js process have changed between 1.39.x and 1.40.x. 1.40.x now includes the parameter --max_old_space_size. Something in the implementation of that is likely the cause.
Even if the tslint extension has a bug, surely having --max_old_space_size=3000 would force it to crash with out of memory instead of using 22Gb of RAM?
I second what @hyarionh is saying, this issues should not be closed.
We had to downgrade all our machines to VSCode to 1.39.x, since this memory leak would freeze the entire OS.
Most helpful comment
This issue should not be closed. The same version of TSLint works perfectly fine on older versions of VSCode which indicates the extension itself is unlikely the cause.
The change might be related to how VSCode starts a new process. The parameters passed to start the tsserver.js process have changed between 1.39.x and 1.40.x. 1.40.x now includes the parameter --max_old_space_size. Something in the implementation of that is likely the cause.
Even if the tslint extension has a bug, surely having --max_old_space_size=3000 would force it to crash with out of memory instead of using 22Gb of RAM?