Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Troubleshooting for Amazon RDS

Use the following sections to help troubleshoot problems you have with DB instances in Amazon RDS and Amazon Aurora.

For information about debugging problems using the Amazon RDS API, see Troubleshooting applications on Amazon RDS.

Can't connect to Amazon RDS DB instance

When you can't connect to a DB instance, the following are common causes:

For engine-specific connection issues, see the following topics:

Testing a connection to a DB instance

You can test your connection to a DB instance using common Linux or Microsoft Windows tools.

From a Linux or Unix terminal, you can test the connection by entering the following. Replace DB-instance-endpoint with the endpoint and port with the port of your DB instance.

nc -zv DB-instance-endpoint port 

For example, the following shows a sample command and the return value.

nc -zv postgresql1.c6c8mn7fake0.us-west-2.rds.amazonaws.com 8299

  Connection to postgresql1.c6c8mn7fake0.us-west-2.rds.amazonaws.com 8299 port [tcp/vvr-data] succeeded! 

Windows users can use Telnet to test the connection to a DB instance. Telnet actions aren't supported other than for testing the connection. If a connection is successful, the action returns no message. If a connection isn't successful, you receive an error message such as the following.

C:\>telnet sg-postgresql1.c6c8mntfake0.us-west-2.rds.amazonaws.com 819

  Connecting To sg-postgresql1.c6c8mntfake0.us-west-2.rds.amazonaws.com...Could not open
  connection to the host, on port 819: Connect failed 

If Telnet actions return success, your security group is properly configured.

Troubleshooting connection authentication

In some cases, you can connect to your DB instance but you get authentication errors. In these cases, you might want to reset the master user password for the DB instance. You can do this by modifying the RDS instance.

For more information about modifying a DB instance, see Modifying an Amazon RDS DB instance.

Amazon RDS security issues

To avoid security issues, never use your master Amazon user name and password for a user account. Best practice is to use your master Amazon Web Services account to create users and assign those to DB user accounts. You can also use your master account to create other user accounts, if necessary.

For information about creating users, see Creating an IAM user in your Amazon Web Services account. For information about creating users in Amazon IAM Identity Center, see Manage identities in IAM Identity Center.

Error message "failed to retrieve account attributes, certain console functions may be impaired."

You can get this error for several reasons. It might be because your account is missing permissions, or your account hasn't been properly set up. If your account is new, you might not have waited for the account to be ready. If this is an existing account, you might lack permissions in your access policies to perform certain actions such as creating a DB instance. To fix the issue, your administrator needs to provide the necessary roles to your account. For more information, see the IAM documentation.

Troubleshooting incompatible-network state

The incompatible-network state means that the database might still be accessible at the database level but you can't modify or reboot it.

Causes

The incompatible-network state of your DB instance could be a result of one of the following actions:

Resolution

Use start-db-instance command

To fix a database that is in an incompatible-network state, follow these instructions:

Your database status changes to Available if the command executes successfully.

If your database restarts, the DB instance might execute the last operation run on the instance before it was moved to incompatible-network state. This might move the instance back to the incompatible-network state.

If the start-db-instance command is unsuccessful or the instance moves back to incompatible-network state, open the Databases page in the RDS console and select the database. Navigate to the Logs & events section. The Recent events section displays further resolution steps to follow. The messages are classified as follows:

Perform point-in-time recovery

It is best practice to have a backup (snapshot or logical), in case your database enters incompatible-network state. See Working with backups. If you turned on automated backups, then temporarily stop any writes to the database and perform a point-in-time recovery.

If you did't turn on automated backups, create a new DB instance. Then migrate the data using Amazon Database Migration Service (Amazon DMS), or by using a backup and restore tool.

If this does not resolve the issue, contact Amazon Web Services Support for further assistance.

Resetting the DB instance owner password

If you get locked out of your DB instance, you can log in as the master user. Then you can reset the credentials for other administrative users or roles. If you can't log in as the master user, the Amazon account owner can reset the master user password. For details of which administrative accounts or roles you might need to reset, see Master user account privileges.

You can change the DB instance password by using the Amazon RDS console, the Amazon CLI command modify-db-instance, or by using the ModifyDBInstance API operation. For more information about modifying a DB instance, see Modifying an Amazon RDS DB instance.

Amazon RDS DB instance outage or reboot

A DB instance outage can occur when a DB instance is rebooted. It can also occur when the DB instance is put into a state that prevents access to it, and when the database is restarted. A reboot can occur when you manually reboot your DB instance. A reboot can also occur when you change a DB instance setting that requires a reboot before it can take effect.

A DB instance reboot occurs when you change a setting that requires a reboot, or when you manually cause a reboot. A reboot can occur immediately if you change a setting and request that the change take effect immediately. Or it can occur during the DB instance's maintenance window.

A DB instance reboot occurs immediately when one of the following occurs:

A DB instance reboot occurs during the maintenance window when one of the following occurs:

When you change a static parameter in a DB parameter group, the change doesn't take effect until the DB instance associated with the parameter group is rebooted. The change requires a manual reboot. The DB instance isn't automatically rebooted during the maintenance window.

To see a table that shows DB instance actions and the effect that setting the Apply Immediately value has, see Modifying an Amazon RDS DB instance.

Amazon RDS DB parameter changes not taking effect

In some cases, you might change a parameter in a DB parameter group but don't see the changes take effect. If so, you likely need to reboot the DB instance associated with the DB parameter group. When you change a dynamic parameter, the change takes effect immediately. When you change a static parameter, the change doesn't take effect until you reboot the DB instance associated with the parameter group.

You can reboot a DB instance using the RDS console. Or you can explicitly call the RebootDBInstance API operation. You can reboot without failover if the DB instance is in a Multi-AZ deployment. The requirement to reboot the associated DB instance after a static parameter change helps mitigate the risk of a parameter misconfiguration affecting an API call. An example of this is calling ModifyDBInstance to change the DB instance class. For more information, see Modifying parameters in a DB parameter group.

Amazon RDS DB instance running out of storage

If your DB instance runs out of storage space, it might no longer be available. We highly recommend that you constantly monitor the FreeStorageSpace metric published in CloudWatch to make sure that your DB instance has enough free storage space.

If your database instance runs out of storage, its status changes to storage-full. For example, a call to the DescribeDBInstances API operation for a DB instance that has used up its storage outputs the following.

aws rds describe-db-instances --db-instance-identifier mydbinstance

DBINSTANCE  mydbinstance  2009-12-22T23:06:11.915Z  db.m5.large  mysql8.0  50  sa
storage-full  mydbinstance.clla4j4jgyph.us-east-1.rds.amazonaws.com  3306
us-east-1b  3
	SECGROUP  default  active
	PARAMGRP  default.mysql8.0  in-sync

To recover from this scenario, add more storage space to your instance using the ModifyDBInstance API operation or the following Amazon CLI command.

For Linux, macOS, or Unix:

aws rds modify-db-instance \
    --db-instance-identifier mydbinstance \
    --allocated-storage 60 \
    --apply-immediately

For Windows:

aws rds modify-db-instance ^
    --db-instance-identifier mydbinstance ^
    --allocated-storage 60 ^
    --apply-immediately

DBINSTANCE  mydbinstance  2009-12-22T23:06:11.915Z  db.m5.large  mysql8.0  50  sa
storage-full  mydbinstance.clla4j4jgyph.us-east-1.rds.amazonaws.com  3306
us-east-1b  3  60
	SECGROUP  default  active
	PARAMGRP  default.mysql8.0  in-sync

Now, when you describe your DB instance, you see that your DB instance has modifying status, which indicates the storage is being scaled.

aws rds describe-db-instances --db-instance-identifier mydbinstance 

DBINSTANCE  mydbinstance  2009-12-22T23:06:11.915Z  db.m5.large  mysql8.0  50  sa
modifying  mydbinstance.clla4j4jgyph.us-east-1.rds.amazonaws.com
3306  us-east-1b  3  60
	SECGROUP  default  active
	PARAMGRP  default.mysql8.0  in-sync

After storage scaling is complete, your DB instance status changes to available.

aws rds describe-db-instances --db-instance-identifier mydbinstance 

DBINSTANCE  mydbinstance  2009-12-22T23:06:11.915Z  db.m5.large  mysql8.0  60  sa
available  mydbinstance.clla4j4jgyph.us-east-1.rds.amazonaws.com  3306
us-east-1b  3
	SECGROUP  default  active
	PARAMGRP  default.mysql8.0  in-sync

You can receive notifications when your storage space is exhausted using the DescribeEvents operation. For example, in this scenario, if you make a DescribeEvents call after these operations you see the following output.

aws rds describe-events --source-type db-instance --source-identifier mydbinstance 

2009-12-22T23:44:14.374Z  mydbinstance  Allocated storage has been exhausted db-instance
2009-12-23T00:14:02.737Z  mydbinstance  Applying modification to allocated storage db-instance
2009-12-23T00:31:54.764Z  mydbinstance  Finished applying modification to allocated storage

Amazon RDS insufficient DB instance capacity

The InsufficientDBInstanceCapacity error can be returned when you try to create, start, or modify a DB instance. It can also be returned when you try to restore a DB instance from a DB snapshot. When this error is returned, a common cause is that the specific DB instance class isn't available in the requested Availability Zone. You can try one of the following to solve the problem:

For information about troubleshooting instance capacity issues for Amazon EC2, see Insufficient instance capacity in the Amazon EC2 User Guide.

For information about modifying a DB instance, see Modifying an Amazon RDS DB instance.

Freeable memory issues in Amazon RDS

Freeable memory is the total random access memory (RAM) on a DB instance that can be made available to the database engine. It's the sum of the free operating-system (OS) memory and the available buffer and page cache memory. The database engine uses most of the memory on the host, but OS processes also use some RAM. Memory currently allocated to the database engine or used by OS processes isn't included in freeable memory. When the database engine is running out of memory, the DB instance can use the temporary space that is normally used for buffering and caching. As previously mentioned, this temporary space is included in freeable memory.

You use the FreeableMemory metric in Amazon CloudWatch to monitor the freeable memory. For more information, see Overview of monitoring metrics in Amazon RDS.

If your DB instance consistently runs low on freeable memory or uses swap space, consider scaling up to a larger DB instance class. For more information, see DB instance classes.

You can also change the memory settings. For example, on RDS for MySQL, you might adjust the size of the innodb_buffer_pool_size parameter. This parameter is set by default to 75 percent of physical memory. For more MySQL troubleshooting tips, see How can I troubleshoot low freeable memory in an Amazon RDS for MySQL database?

MySQL and MariaDB issues

You can diagnose and correct issues with MySQL and MariaDB DB instances.

Maximum MySQL and MariaDB connections

The maximum number of connections allowed to an RDS for MySQL or RDS for MariaDB DB instance is based on the amount of memory available for its DB instance class. A DB instance class with more memory available results in a larger number of connections available. For more information on DB instance classes, see DB instance classes.

The connection limit for a DB instance is set by default to the maximum for the DB instance class. You can limit the number of concurrent connections to any value up to the maximum number of connections allowed. Use the max_connections parameter in the parameter group for the DB instance. For more information, see Maximum number of database connections and Working with parameter groups.

You can retrieve the maximum number of connections allowed for a MySQL or MariaDB DB instance by running the following query.

SELECT @@max_connections;

You can retrieve the number of active connections to a MySQL or MariaDB DB instance by running the following query.

SHOW STATUS WHERE `variable_name` = 'Threads_connected';

Diagnosing and resolving incompatible parameters status for a memory limit

A MariaDB or MySQL DB instance can be placed in incompatible-parameters status for a memory limit when the following conditions are met:

When a DB instance is restarted for the third time in one hour or for the fifth time in one day, it performs a check for memory usage. The check makes a calculation of the potential memory usage of the DB instance. The value returned by the calculation is the sum of the following values:

So, the value returned by the calculation is the following:

Value 1 + Value 2 + Value 3

When this value exceeds 1.2 times the memory allocated to the DB instance class used by the DB instance, the DB instance is placed in incompatible-parameters status. For information about the memory allocated to DB instance classes, see Hardware specifications for DB instance classes.

The calculation multiplies the value of the max_connections parameter by the sum of several parameters. If the max_connections parameter is set to a large value, it might cause the check to return an inordinately high value for the potential memory usage of the DB instance. In this case, consider lowering the value of the max_connections parameter.

To resolve the problem, complete the following steps:

Diagnosing and resolving lag between read replicas

After you create a MySQL or MariaDB read replica and the replica is available, Amazon RDS first replicates the changes made to the source DB instance from the time the read replica create operation started. During this phase, the replication lag time for the read replica is greater than 0. You can monitor this lag time in Amazon CloudWatch by viewing the Amazon RDS ReplicaLag metric.

The ReplicaLag metric reports the value of the Seconds_Behind_Master field of the MariaDB or MySQL SHOW REPLICA STATUS command. For more information, see SHOW REPLICA STATUS Statement in the MySQL documentation.

When the ReplicaLag metric reaches 0, the replica has caught up to the source DB instance. If the ReplicaLag metric returns -1, replication might not be active. To troubleshoot a replication error, see Diagnosing and resolving a MySQL or MariaDB read replication failure. A ReplicaLag value of -1 can also mean that the Seconds_Behind_Master value can't be determined or is NULL.

The ReplicaLag metric returns -1 during a network outage or when a patch is applied during the maintenance window. In this case, wait for network connectivity to be restored or for the maintenance window to end before you check the ReplicaLag metric again.

The MySQL and MariaDB read replication technology is asynchronous. Thus, you can expect occasional increases for the BinLogDiskUsage metric on the source DB instance and for the ReplicaLag metric on the read replica. For example, consider a situation where a high volume of write operations to the source DB instance occur in parallel. At the same time, write operations to the read replica are serialized using a single I/O thread. Such a situation can lead to a lag between the source instance and read replica.

For more information about read replicas and MySQL, see Replication implementation details in the MySQL documentation. For more information about read replicas and MariaDB, see Replication overview in the MariaDB documentation.

You can reduce the lag between updates to a source DB instance and the subsequent updates to the read replica by doing the following:

Diagnosing and resolving a MySQL or MariaDB read replication failure

Amazon RDS monitors the replication status of your read replicas. RDS updates the Replication State field of the read replica instance to Error if replication stops for any reason. You can review the details of the associated error thrown by the MySQL or MariaDB engines by viewing the Replication Error field. Events that indicate the status of the read replica are also generated, including RDS-EVENT-0045, RDS-EVENT-0046, and RDS-EVENT-0047. For more information about events and subscribing to events, see Working with Amazon RDS event notification. If a MySQL error message is returned, check the error in the MySQL error message documentation. If a MariaDB error message is returned, check the error in the MariaDB error message documentation.

Common situations that can cause replication errors include the following:

The following steps can help resolve your replication error:

If a replication error is fixed, the Replication State changes to replicating. For more information, see Troubleshooting a MySQL read replica problem.

Creating triggers with binary logging enabled requires SUPER privilege

When trying to create triggers in an RDS for MySQL or RDS for MariaDB DB instance, you might receive the following error.

"You do not have the SUPER privilege and binary logging is enabled" 

To use triggers when binary logging is enabled requires the SUPER privilege, which is restricted for RDS for MySQL and RDS for MariaDB DB instances. You can create triggers when binary logging is enabled without the SUPER privilege by setting the log_bin_trust_function_creators parameter to true. To set the log_bin_trust_function_creators to true, create a new DB parameter group or modify an existing DB parameter group.

You can create a new DB parameter group so you can create triggers in your RDS for MySQL or RDS for MariaDB DB instance with binary logging enabled. To do so, use the following CLI commands. To modify an existing parameter group, start with step 2.

Diagnosing and resolving point-in-time restore failures

Restoring a DB Instance That Includes Temporary Tables

When attempting a point-in-time restore (PITR) of your MySQL or MariaDB DB instance, you might encounter the following error.

Database instance could not be restored because there has been incompatible database activity for restore
functionality. Common examples of incompatible activity include using temporary tables, in-memory tables,
or using MyISAM tables. In this case, use of Temporary table was detected.	

PITR relies on both backup snapshots and binary logs (binlogs) from MySQL or MariaDB to restore your DB instance to a particular time. Temporary table information can be unreliable in binlogs and can cause a PITR failure. If you use temporary tables in your MySQL or MariaDB DB instance, you can lower the possibility of a PITR failure. To do so, perform more frequent backups. A PITR failure is most probable in the time between a temporary table's creation and the next backup snapshot.

Restoring a DB Instance That Includes In-Memory Tables

You might encounter a problem when restoring a database that has in-memory tables. In-memory tables are purged during a restart. As a result, your in-memory tables might be empty after a reboot. We recommend that when you use in-memory tables, you architect your solution to handle empty tables in the event of a restart. If you're using in-memory tables with replicated DB instances, you might need to recreate the read replicas after a restart. This might be necessary if a read replica reboots and can't restore data from an empty in-memory table.

For more information about backups and PITR, see Working with backups and Restoring a DB instance to a specified time.

Replication stopped error

When you call the mysql.rds_skip_repl_error command, you might receive an error message stating that replication is down or disabled.

This error message appears because replication is stopped and can't be restarted.

If you need to skip a large number of errors, the replication lag can increase beyond the default retention period for binary log files. In this case, you might encounter a fatal error due to binary log files being purged before they have been replayed on the replica. This purge causes replication to stop, and you can no longer call the mysql.rds_skip_repl_error command to skip replication errors.

You can mitigate this issue by increasing the number of hours that binary log files are retained on your replication source. After you have increased the binlog retention time, you can restart replication and call the mysql.rds_skip_repl_error command as needed.

To set the binlog retention time, use the mysql.rds_set_configuration procedure. Specify a configuration parameter of 'binlog retention hours' along with the number of hours to retain binlog files on the DB cluster, up to 720 (30 days). The following example sets the retention period for binlog files to 48 hours.

CALL mysql.rds_set_configuration('binlog retention hours', 48);

Read replica create fails or replication breaks with fatal error 1236

After changing default parameter values for a MySQL or MariaDB DB instance, you might encounter one of the following problems:

Some default parameter values for MySQL and MariaDB DB instances help to make sure that the database is ACID compliant and read replicas are crash-safe. They do this by making sure that each commit is fully synchronized by writing the transaction to the binary log before it's committed. Changing these parameters from their default values to improve performance can cause replication to fail when a transaction hasn't been written to the binary log.

To resolve this issue, set the following parameter values:

Can't set backup retention period to 0

There are several reasons why you might need to set the backup retention period to 0. For example, you can disable automatic backups immediately by setting the retention period to 0.

In some cases, you might set the value to 0 and receive a message saying that the retention period must be between 1 and 35. In these cases, check to make sure that you haven't set up a read replica for the instance. Read replicas require backups for managing read replica logs, and therefore you can't set a retention period of 0.