Error starting rds instance RDS:xxx (Shape is missing required key 'type': OrderedDict([('members', OrderedDict([('SourceRegion', OrderedDict([('shape', 'String')

RDS instances are not starting after stopping successfully with the Instance Scheduler:

2021-03-26 - 08:00:28.186 - ERROR   : Error starting rds instance RDS:xxx
(Shape is missing required key 'type': OrderedDict([('members', OrderedDict([('SourceRegion', OrderedDict([('shape', 'String'),
('documentation', '<p>The ID of the region that contains the source for the db instance.</p>')]))]))]))

We are using the aws-instance-scheduler version v1.3.3 (SO0030).

CloudFormation parameters:

• DefaultTimezone: America/Sao_Paulo
• LogRetentionDays: 1
• MemorySize: 1024
• Regions: us-east-2
• ScheduleLambdaAccount: Yes
• ScheduleRdsClusters: Yes
• ScheduledServices: Both
• SchedulerFrequency: 1
• SchedulingActive: Yes
• SendAnonymousData: Yes
• StartedTags: -
• StoppedTags: -
• TagName: Schedule
• Trace: Yes
• UseCloudWatchMetrics: Yes

We created the period and schedule using the scheduler-cli tool running the following commands:

scheduler-cli create-period --name cwb-office-hours --weekdays mon-fri --begintime 06:00 --endtime 20:00 --stack Instance-Scheduler
scheduler-cli create-schedule --name cwb-office-hours --periods cwb-office-hours --stack Instance-Scheduler

DynamoDB ConfigTable:

We are using the tag Schedule with the value cwb-office-hours with RDS Instances (MySQL, PostgreSQL) and RDS Clusters (Aurora MySQL).

[esaporski]

It worked 😃 I checked the logs and it is behaving a little bit erratically though:

2021-04-06 - 09:00:32.886 - DEBUG   : Desired state for instance from schedule "cwb-office-hours" is stopped,
last desired state was unknown, actual state is running 

Thanks a lot, @gockle!

[gockle]

@esaporski Is it possible to update the lambda code and try a workaround? the file rds_service.py in the ./lambda/schedulers folder, update the method start_instances(…) by removing the return statement at the end. This is a know issue, when there is an error in one of the instances the code exits the method without completing the start for all the instances in the list. We will be adding a change to our next release to address this issue, but let me know if the workaround resolves the issue and starts all the instances.