Swarming: Avoids mounting volume

[IvanBM18]

Overview

Swarming, as well as K8S has issues when trying to mount this volume to the container.

mount /mnt/scratch0 -o remount,exec,suid,dev

So we added a validation to not mount when we are either in K8S or in swarming.

Note: Uses SWARMING_BOT env var as its the one we already use in the swarming request

Tests performed

I created a custom bash script to test this:

is_truthy() {
  [[ "$1" =~ ^([Tt]rue|1)$ ]]
}

should_mount() {
  if is_truthy "$IS_K8S_ENV" || is_truthy "$SWARMING_BOT"; then
    return 1
  fi
  return 0
}

test_mount() {
  unset IS_K8S_ENV SWARMING_BOT
  [[ "$1" != "unset" ]] && export IS_K8S_ENV=$1
  [[ "$2" != "unset" ]] && export SWARMING_BOT=$2
  printf "IS_K8S_ENV=%-5s | SWARMING_BOT=%-5s  =>" "$1" "$2"
  if should_mount; then echo " [MOUNT EXECUTED]"; else echo " [MOUNT SKIPPED]"; fi
}
'

Against multiple inputs:

echo "=== DRY RUN OUTPUT USING is_truthy() ==="
test_mount "unset" "unset"
test_mount "1"     "unset"
test_mount "true"  "unset"
test_mount "false" "unset"
test_mount "False" "unset"
test_mount "0"     "unset"
test_mount "unset" "True"
test_mount "unset" "1"
test_mount "unset" "false"
test_mount "false" "false"

And got this results:

=== DRY RUN OUTPUT USING is_truthy() ===
IS_K8S_ENV=unset | SWARMING_BOT=unset  => [MOUNT EXECUTED]
IS_K8S_ENV=1     | SWARMING_BOT=unset  => [MOUNT SKIPPED]
IS_K8S_ENV=true  | SWARMING_BOT=unset  => [MOUNT SKIPPED]
IS_K8S_ENV=false | SWARMING_BOT=unset  => [MOUNT EXECUTED]
IS_K8S_ENV=False | SWARMING_BOT=unset  => [MOUNT EXECUTED]
IS_K8S_ENV=0     | SWARMING_BOT=unset  => [MOUNT EXECUTED]
IS_K8S_ENV=unset | SWARMING_BOT=True   => [MOUNT SKIPPED]
IS_K8S_ENV=unset | SWARMING_BOT=1      => [MOUNT SKIPPED]
IS_K8S_ENV=unset | SWARMING_BOT=false  => [MOUNT EXECUTED]
IS_K8S_ENV=false | SWARMING_BOT=false  => [MOUNT EXECUTED]

[jardondiego]

We have a failing check, is that failure related to this change?

[javanlacerda]

/gcbrun

[javanlacerda]

LGTM. I'm just afraid we would need to generate a new base image and update it for the clusterfuzz instances for applying this change, and it seems the pipeline for building the base images is broken.

You think running the swarming containers with IS_K8S_ENV=True would mitigate it for now?

[IvanBM18]

LGTM. I'm just afraid we would need to generate a new base image and update it for the clusterfuzz instances for applying this change, and it seems the pipeline for building the base images is broken.

You think running the swarming containers with IS_K8S_ENV=True would mitigate it for now?

Sure we can work with that till the main issue is fixed