Sometimes the lambda tries to tag snapshots that have not yet been created.

[legkovalex]

Retaining snapshot snap-071f711539ba0276a of volume vol-a139336bed from instance i-9117b492be for 7 days
Will delete 9 snapshots on 2017-12-14
An error occurred (InvalidSnapshot.NotFound) when calling the CreateTags operation: The snapshot 'snap-071f711539ba0276a' does not exist.: ClientError
Traceback (most recent call last):

Sometimes snapshots do not have time to be created before they start tagging. It seems that a small timeout between the creation of snapshots and their tagging will not be superfluous.

[cmachler]

Just curious how many snapshots are trying to be created in your environment?

[legkovalex]

Approximately ten snapshots in three regions. The error is not constant and occurs once for a dozen times, approximately. The most unpleasant thing is that we must clear snapshots manually since they are not tagged for deletion (within the region).
I think that the possibility of setting the timeout would be appropriate. Probably, as a variable (such as regions) with a default value of 0 seconds (if empty). With the possibility of increasing if necessary.

[cmachler]

I think that sounds reasonable with regards to the timeout variable @legkovalex.

Would you have time to work on this feature and put in a PR? I do not have the time currently to work on this in a expedited manor.