agrrh/getlock-docs repository overview

⁠Info

GetLock.tech⁠ service exists for those who need to get some kind of session lock or simple semaphore-like mechanism.

Actual case was absence of proper concurrency limit for BitBucket Pipelines⁠.

⁠Usage

⁠PUT https://getlock.tech/v1/<uuid>

Create new lock or update existing one.

Params:

• ttl - float, in seconds, default is 60.0.

Response codes:

• 201 created
• 200 refreshed

⁠GET https://getlock.tech/v1/<uuid>

Check if lock is active.

Response codes:

• 423 for active
• 404 for other cases

⁠DELETE https://getlock.tech/v1/<uuid>

Release existing lock.

Response codes:

• 200 success
• 404 lock not found

⁠Example

Let's say we have some task which takes up to 1 hour yet could be triggered from number of workers. It concurrency required to be limited to single worker at a time.

We choose some random yet valid UUID⁠ as our job ID. It should be the same for all workers you decide to limit concurrency for:

YOUR_JOB_UUID=28c9c1be-7bb8-4e8c-b22e-c16402a421d5

Worker logic then could look like following:

echo "Checking if another job is running ..."

while [ "$(curl \"https://getlock.tech/v1/${YOUR_JOB_UUID}\" | jq '.locked')" = "true" ]; do
  echo "Locked, waiting for free spot ..."
  sleep 1
done

echo "Free spot found"

echo "Acquiring lock ..."

# NOTE Prefer lower TTL values while refreshing lock periodically
#   to not waste time in case job has failed
curl -X PUT -d ttl=3600 "https://getlock.tech/v1/${YOUR_JOB_UUID}"

echo "Starting payload ..."

sleep 2400

echo "Payload done"

echo "Releasing lock ..."

curl -X DELETE "https://getlock.tech/v1/${YOUR_JOB_UUID}"

echo "Job done"

I also recommend you to consider using awesome httpie⁠ tool.

⁠Credits

• Consul Sessions⁠ - inspiration
• agrrh⁠ - implementation