Events when a temporarily stopped RDS Instance is restarted

0

I have several RDS instances, all of them MariaDB 10.6.10, single instances, no multi-az deployment. These instances are temporarily stopped all days at evening and restarted on demand. Every time one of them is restarted we get this events (I'm including the oldest event, the stopping of the instance):

TimeSystem notes
October 27, 2022, 11:35 (UTC+02:00)Finished updating DB parameter group
October 27, 2022, 11:34 (UTC+02:00)Monitoring Interval changed to 0
October 27, 2022, 11:34 (UTC+02:00)Performance Insights has been enabled
October 27, 2022, 11:33 (UTC+02:00)DB instance started
October 27, 2022, 11:31 (UTC+02:00)Recovery of the DB instance is complete.
October 27, 2022, 11:31 (UTC+02:00)DB instance restarted
October 27, 2022, 11:31 (UTC+02:00)DB instance restarted
October 27, 2022, 11:30 (UTC+02:00)Recovery of the DB instance has started. Recovery time will vary with the amount of data to be recovered.
October 27, 2022, 11:29 (UTC+02:00)Recovery of the DB instance has started. Recovery time will vary with the amount of data to be recovered.
October 26, 2022, 19:08 (UTC+02:00)DB instance stopped

It doesn't matter if the instance has data on it or is freshly created with no data at all and immediately stopped. Every time is restarted we get the events on the table..., so here my questions:

  • Why "Recovery of the DB instance has started"? It sounds to me as something is wrong with the instance: data corruption, system crash... but system crash didn't happen, even with new instances recently created and stopped the "Recovery of the DB instance has started" happens, and it could not be that all instances have corrupted data... So, is this event part the normal behaviour? Is it happening to every one in every instance? I couldn't find too much info about this event https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_Events.Messages.html#USER_Events.Messages.instance and in no place is told that in a RDS restart this event will happen...
  • And other question, why do some events happen / are logged twice?

Thank you in advance for your help, Marcos

asked a year ago714 views
3 Answers
1

Database recovery is performed on every startup even after a clean shutdown. This is quite normal and not at all an issue.

AWS
MODERATOR
philaws
answered a year ago
  • It's a relief that someone at AWS states that is quiet normal and not an issue! Anyway, it would be great that it was stated in the documentation as well. And why these events are registered twice?

0

Database recovery is a normal way to make sure that the database log files are correctly applied or rolled back to maintain transactional correctness.

My guess is that when you stop an RDS instance it doesn't wait for all transactions to finish (as this could cause a delay in the stopping of the database.). Instead it immediately kills all connections and stops the database. This may leave the database in a state where there are committed items in the logs that still need to be applied or uncommitted items in the logs that need to be rolled back in order to get the database into a proper transactional running state.

answered a year ago
0

Thank you for your answer @joelrwilliams1! So, your thought , like mine, is that the database is not on a "stable" state, that something went wrong when the RDS instance was stopped. Well, I'm agree, but I'm trying to find out what / who is the cause... And I can assure it's not my fault, because I take care to have not open connections to the instance when I stop the database, it happens every time with every instance, even with new created instances: I create it, I don't tell anyone the host name, user or password, even the security group doesn't allow anyone to access to the mariadb instance, so no user connections exist, I wait for the instance to be in Avalaible state ... and then in actions I click on "Stop temporarily" ... and yet, when I restart the Instance the "Recovery of the DB instance has started" happens, so I'm afraid that using this feature are endangering my database and my data. Could anyone else try to reproduce this behaviour? As my point of view, is AWS who are not stopping database in an orderly and gentle way ... Could anyone from AWS shed any light on this matter? Thanks again and best regards

answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions