Hi, what is the right way to cope with a locked st...
# getting-started
Hi, what is the right way to cope with a locked stack? How can I distinguish between a case that a stack is locked because another process is currently updating it, and the case when the CLI process got killed while performing an update? In the first case I would like to do nothing (or allow update only after a certain time period has passed), and in the second case I would like to unlock the stack and retry the update
Some background: I am working with the automation API with an inline program, currently with local backend. Currently, I deal with a locked stack in the following way: • when calling
, if I get a stack locked err message I run
• then I get another error when calling
saying there are pending resources for the stack • I then call
• then
finally works I am not sure this is the right way to cope with it, especially as I don't have a way to distinguish between the 2 cases as mentioned above. Also, the way I determine the error type is by textual search on the error message which is far from optimal...
I don't think you can distinguish between them currently. You just have to know what processes may be running to update the stack and see if any of them are currently running. We've had ideas to add timers or heartbeats or something to keep stacks locked to try and help with this, but nothing planned out for sure yet.
Is this the case also when working with Pulumi service backend? Or just file based backend?
Also, is there a way to get a timestamp of when the lock was set?
And, if working with the Pulumi service, does the lock get released after some time automatically?
I'm not 100% sure I'll have to check with the service team but I don't think those locks are release automatically either. You might be able to get the created timestamp from the lock file in the file state folder/bucket, I think the path will be something like ".pulumi/locks/<stack name>/random-uuid.json"
Thanks. If you can check with the service guys that will really help
Posted internally. All our service team is US based so probably won't get an answer till tonight though.
🙏 1
The service locks should timeout after 5 minutes of inactivity.
Great, thanks!