v20240829-190428

Allow setting minimum available runners to 0 (#5597)

This change modifies the check to allow the minimum available runners
setting to be 0 while keeping the default value to 10. The sanity check
to ensure that at least a positive number is used remains. If an invalid
number is passed then the setting is assumed to be 0.

Scale-up function however requires a minimum of at least 1 to be able
to do its calculations on how many runners to scale up. A value of 0
here breaks the logic and causes the function to never create any
runners.
So logic is added here to ensure that scale-up's min value is 1.

Issue: https://lf-pytorch.atlassian.net/browse/PC-33

Signed-off-by: Thanh Ha <[email protected]>

v20240826-204017

Enable setup minimum available runners for scale infra (#5594)

adds `min_available_runners` terraform config to setup environment
variable `MIN_AVAILABLE_RUNNERS` on scale lambdas

v20240821-203210

fix: TFLint error for duplicate output declaration (#5580)

The output here is duplicated and is causing TFLint to throw an error.

Signed-off-by: Thanh Ha <[email protected]>

v20240821-203158

fix: Download ARM64 binary (#5577)

The lambda runner-binaries-syncer is missing code to actually perform
the
download for the ARM64 binary despite having the variables in place.
This change resolves that issue so that the syncer actually fetches
the appropriate binary for ARM64.

Closes: pytorch/ci-infra#263
Depends-On: #5569

---------

Signed-off-by: Thanh Ha <[email protected]>

v20240821-203008

fix: runner-binaries-syncer missing requests module (#5569)

The runner-binaries-syncer fails with missing module "requests". Seems
like in the past it worked fine without it but is now showing up as an
error.

Relates to pytorch/ci-infra#263.

Signed-off-by: Thanh Ha <[email protected]>

v20240816-181343

Remove default for AMIs and add ami_filter_linux_arm64 ami_filter_lin…

v20240807-193314

Runner variants to be aware of the lf. and lf.c. prefixes, to facilit…

v20240802-160133

Introduce runner variants to help test changes (#5531)

v20240724-192912

Fix bug with emitting logs from runners (#5503)

v20240724-185114

On ScaleDown remove old/stale SSM parameters (#5479)