I have many GAE services that are running on GAE Flex. The services are configured to run with minimum 1 instance and maximum 5 instances. Today, the services are automatically down due to the shutdown script triggered by system. Based on the google support's investigation, the shutdown triggered was because of the upgrade performed on the instance. Before the restart, there is only 1 running instance for the GAE services, which resulted that this service is down for a period of time. Also for the restart for upgrade, this does not trigger to start another instance for the GAE services. As a result, the support suggested to have 2 or 3 running instances for the GAE services at all time to avoid downtime. This suggestion will increase the running cost by 2-3 times for everyday, even we don't need so many instances to be running at all time. Also this is no eco-friendly to run so many instances all the time when it is not needed.
What you would like to accomplish: I would like to avoid having downtime during the upgrade period of the instance triggered by Google. GAE is built to automatically manage the load for the application.
How this might work: - Before the start of the upgrade, Google can check if there is only 1 instance running and does the config allows more instances to be created. - Start the instance if condition fulfilled. - Once instance is started, Google can proceed to upgrade the instance. - After upgrade and restart is completed, the instance can adjust back to the original configuration.
If applicable, reasons why alternative solutions are not sufficient: As mentioned above, the cost incurred by running 2-3 instances at all time is 2-3 times more. It defeat the purpose to use GAE. Lastly, this is not being eco-friendly to have 2-3 instances running at all time.
Other information (workarounds you have tried, documentation consulted, etc):
Description
Today, the services are automatically down due to the shutdown script triggered by system. Based on the google support's investigation, the shutdown triggered was because of the upgrade performed on the instance.
Before the restart, there is only 1 running instance for the GAE services, which resulted that this service is down for a period of time.
Also for the restart for upgrade, this does not trigger to start another instance for the GAE services.
As a result, the support suggested to have 2 or 3 running instances for the GAE services at all time to avoid downtime. This suggestion will increase the running cost by 2-3 times for everyday, even we don't need so many instances to be running at all time.
Also this is no eco-friendly to run so many instances all the time when it is not needed.
What you would like to accomplish:
I would like to avoid having downtime during the upgrade period of the instance triggered by Google.
GAE is built to automatically manage the load for the application.
How this might work:
- Before the start of the upgrade, Google can check if there is only 1 instance running and does the config allows more instances to be created.
- Start the instance if condition fulfilled.
- Once instance is started, Google can proceed to upgrade the instance.
- After upgrade and restart is completed, the instance can adjust back to the original configuration.
If applicable, reasons why alternative solutions are not sufficient:
As mentioned above, the cost incurred by running 2-3 instances at all time is 2-3 times more.
It defeat the purpose to use GAE.
Lastly, this is not being eco-friendly to have 2-3 instances running at all time.
Other information (workarounds you have tried, documentation consulted, etc):