Versions Involved

Version 3.1.0.4 or 3.1.0.5

Version 3.0.8

Upgrade Reason

In occasional scenarios, MySQL clients cannot terminate queries. We suggest you upgrade the version to 3.1.0.7 during off-peak hours.

If your node is slow to start when there is a large number of failed distributed transactions, we suggest you upgrade the version to 3.0.8.5 during off-peak hours.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/usermanual-ddm/ddm_06_0042.html

Estimated Interruption

The interruption is no more than 60s for regular workloads.

Estimated Upgrade Duration

The duration depends on how many instance nodes you have. Upgrading a single node takes about 5 minutes.

Upgrade Impacts

DDM instances are upgraded using rolling upgrades for DDM nodes. There will be brief interruptions (less than 1 minute) that the applications can detect.

To minimize the impact, perform the upgrade during off-peak hours and ensure that your applications support automatic reconnection.

For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/usermanual-ddm/ddm_06_0042.html

Versions Involved

Version 2.22.07.000

Version 2.23.06.000, 2.23.06.001, or 2.23.09.000

Upgrade Reason

A database proxy of an earlier version may cause high CPU usage in specific scenarios, which may cause performance risks. To minimize the impact, we suggest you upgrade the database proxy to the latest version during off-peak hours.

The current database proxy version has a bug: During the Prepare phase, an error occasionally occurs while the SELECT FOR UPDATE statement is being executed. We suggest you upgrade the database proxy to the latest version during off-peak hours.

Upgrade Instructions

1. RDS for MySQL database proxy

2. GaussDB(for MySQL) database proxy

https://support.huaweicloud.com/intl/en-us/usermanual-taurusdb/taurusdb_11_0044.html

Estimated Interruption

1. For the database proxy working in clustering mode

Persistent connections that have been active for more than 24 hours will be briefly disconnected (no more than 10s).

2. For the database proxy working in primary/standby mode

The interruption is no more than 10s for regular workloads.

Estimated Upgrade Duration

About 10 min

Upgrade Impacts

1. For the database proxy working in clustering mode

The system will create proxy nodes of the new version during the upgrade. After the nodes are created, applications establish new connections to the new proxy nodes. The proxy nodes of the old version continue to run for 24 hours and then disable any remaining persistent connections.

Perform the upgrade during off-peak hours and ensure that your application supports automatic reconnection.

2. For the database proxy working in primary/standby mode

Workloads are interrupted for a few seconds during the upgrade. Perform the upgrade during off-peak hours and ensure that your application supports automatic reconnection.

3.  For details about the upgrade operations and impacts, visit the following websites:

RDS for MySQL database proxy

GaussDB(for MySQL) database proxy

https://support.huaweicloud.com/intl/en-us/usermanual-taurusdb/taurusdb_11_0044.html

Versions Involved

2.0.28.15  2.0.28.16   2.0.28.17   2.0.29.1   2.0.29.2

Upgrade Reason

GaussDB(for MySQL) has recently resolved many issues raised by the MySQL open-source community. To ensure service stability, we suggest you upgrade the minor version of your GaussDB(for MySQL) instance during off-peak hours.

Upgrade Instructions

 https://support.huaweicloud.com/intl/en-us/usermanual-taurusdb/taurusdb_05_2265.html

Estimated Interruption

An interruption is no more than 10s for regular workloads.

Estimated Upgrade Duration

The upgrade duration depends on how many instance nodes you have. Upgrading a single node takes about 5 minutes.

Upgrade Impacts

1. A minor version upgrade will cause a GaussDB(for MySQL) instance to reboot and briefly interrupt services. To minimize the impact of the upgrade, perform the upgrade during off-peak hours and ensure that your applications support automatic reconnection.

2. If the primary node is in the same AZ as the read replicas of a DB instance, a minor version upgrade will trigger a switchover. If they are in different AZs, a minor version upgrade will trigger two switchovers. A switchover means that workloads of the primary node are switched over to a read replica because the primary node is unavailable.

3. For details about the upgrade operations and impacts, visit  https://support.huaweicloud.com/intl/en-us/usermanual-taurusdb/taurusdb_05_2265.html

Versions Involved

Version 8.0.25

Version 5.7.27

Version 5.7.21

Upgrade Reason

In MySQL 8.0.25 (patch version: 8.0.25.1), sessions waiting for metadata locking (MDL) lock are counted as active sessions. When an instance performs Data Definition Language (DDL) operations on a table, the instance holds the MDL lock and other Data Manipulation Language (DML) sessions updating this table need to wait for the lock to be released. As a result, there are too many active threads in the database, and new database connections cannot be established. This issue has been resolved in the latest version. We suggest you upgrade the minor version to the latest version during off-peak hours. If you have any questions, submit a service ticket.

In MySQL 5.7.27 (patch versions: 5.7.27.1 and 5.7.27.4), new connections may fail to be established and the DB instance status is abnormal due to early version bugs. This issue has been resolved in the latest version. We suggest you upgrade the minor version to the latest version during off-peak hours. If you have any questions, submit a service ticket.

MySQL 5.7.21 does not support audit logging. If this function is needed, upgrade the minor version to the latest version during off-peak hours.

Upgrade Instructions

Estimated Interruption

1. Single-node instance (or read replica)

The interruption is no more than 5 min for regular workloads.

2. Primary/standby instance

The interruption is no more than 60s for regular workloads.

Estimated Upgrade Duration

Single-node instance (or read replica): about 7 min

Primary/standby instance: about 15 min

Upgrade Impacts

1. A minor version upgrade will cause an RDS for MySQL instance to reboot and briefly interrupt services. To minimize the impact of the upgrade, perform the upgrade during off-peak hours and ensure that your applications support automatic reconnection.

2· If primary and standby DB instances are in the same AZ, a minor version upgrade will trigger a switchover. If they are in different AZs, a minor version upgrade will trigger two switchovers.

2. When you upgrade a minor version of the primary instance, minor versions of read replicas (if any) will also be upgraded. Perform the upgrade during off-peak hours because the DB instance will be rebooted after the upgrade is complete. (Minor versions of read replicas cannot be upgraded separately.)

3. During a kernel version upgrade, if the RDS instance is the source from which to replicate data, DRS may fail to obtain binlogs of this instance. If the RDS instance is the destination of data replication, DRS may fail to write data to the destination database.

4. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/usermanual-rds-mysql/rds_05_0003.html

Versions Involved

Enhanced Edition

Upgrade Reason

The enhanced edition is no longer sold. For instances created from 2020/10/1 to 2021/06/30, when database parameters are modified, the database may restart. This problem has been resolved in the latest version. We suggest you upgrade the minor version to the latest version during off-peak hours. If you have any questions, submit a service ticket.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/usermanual-rds-mysql/rds_05_0003.html

Estimated Interruption

The upgrade causes instances to reboot. There will be an interruption lasting about 5-10 min.

Estimated Upgrade Duration

The upgrade takes 5-10 min for regular workloads.

Upgrade Impacts

1. The upgrade takes about 5-10 min when no workloads are running, but if there is a lot of data, it will take longer. Perform the upgrade during off-peak hours and ensure that your application supports automatic reconnection. If the instance has read replicas, the minor version of the read replicas will also be upgraded.

2. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/usermanual-rds-mysql/rds_05_0003.html

Versions Involved

MongoDB 3.4 or 4.0

MongoDB 4.2

Upgrade Reason

In MongoDB 3.4 (patch version: 3.4.14.7) and 4.0 (patch version: 4.0.3.8), slow query logs occasionally fail to be reported.

We suggest you upgrade the minor version to the latest version during off-peak hours. If you have any questions, submit a service ticket.

Open-source community issues of MongoDB 4.2 (patch version: 4.2.0.230530) may cause session tables to fail to be updated. When session objects are stacked in the memory, the connection response becomes slow (https://jira.mongodb.org/browse/SERVER-39044 ).

Upgrade the minor version to the latest version during off-peak hours. If you have any questions, submit a service ticket.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/usermanual-dds/dds_03_0211.html

Estimated Interruption

1. If the DB instance type is replica set, an interruption is less than 30s.

2. If the DB instance type is cluster, each shard is briefly interrupted (less than 30s) during the upgrade.

Estimated Upgrade Duration

1. For the DB instance type replica set

About 20 minutes.

2. For the DB instance type cluster

Upgrading one shard takes 9-15 min on average. The total duration of upgrading a cluster instance equals to the accumulated result of the upgrade time of all shards.

Upgrade Impacts

1.The upgrade will cause the instance to reboot and interrupt services (one time, for less than 30 seconds). Perform the upgrade during off-peak hours and ensure that your application supports automatic reconnection.

2. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/usermanual-dds/dds_03_0211.html

Versions Involved

1.7

Upgrade Reason

The version 1.7 (patch versions: 1.7.4.3, 1.7.4.5, and 1.7.4.10) has known issues. Upgrade it to the latest version during off-peak hours so your instance can have the functions, O&M capabilities, and enhanced performance and stability the latest version provides. If you have any questions, submit a service ticket.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/influxug-nosql/nosql_02_0733.html

Estimated Interruption

A single node is interrupted for no more than 10s. The actual duration and number of interruptions depend on how many instance nodes you have.

Estimated Upgrade Duration

The upgrade takes about 30 min. The actual duration depends on how many instance nodes you have.

Upgrade Impacts

1. The instance will be restarted and services may be interrupted during the upgrade. The interruption duration depends on workloads, number of nodes, and data volume. Upgrade your instance during off-peak hours and ensure that your application supports automatic reconnection.

2. Upgrading the minor version of an instance will restart each node of the instance in sequence. When a node is being restarted, its services will be taken over by another node. Each takeover will interrupt services for 5s to 10s. The upgrade lasts about half an hour.

3. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/influxug-nosql/nosql_02_0733.html

Versions Involved

3.11.3

Upgrade Reason

The version 3.11.3 (patch versions: 3.11.3.46, 3.11.3.50, 3.11.3.67, 3.11.3.64, and 3.11.3.68) has known issues. Upgrade it to the latest version during off-peak hours so your instance can have the functions, O&M capabilities, and enhanced performance and stability the latest version provides. If you have any questions, submit a service ticket.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/cassandraug-nosql/nosql_02_0731.html

Estimated Interruption

When you upgrade a cluster, services may be interrupted a number of times equal to the number of nodes in the cluster plus one. Each interruption will last for no more than 60s and will only affect the services on that node.

Estimated Upgrade Duration

You can see the user guide to see details about the upgrade duration. For example, if there are three nodes in a cluster instance, the upgrade takes 13-16 min.

Upgrade Impacts

1. The instance will be restarted and services may be interrupted during the upgrade. The length of the interruption depends on the workloads, the number of nodes, and how much data there is. Upgrade your instance during off-peak hours and ensure that your application supports automatic reconnection.

2. When you upgrade a cluster, services may be interrupted a number of times equal to the number of nodes in the cluster plus one. Each interruption will last for no more than 60s and will only affect the services on that node.

3. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/cassandraug-nosql/nosql_02_0731.html

Versions Involved

5.0

Upgrade Reason

The version 5.0 (instances using the patch versions created during 2023/2/1 to 2023/4/30 and 2023/9/10 to 2023/9/20) has known issues. Upgrade it to the latest version during off-peak hours so your instance can have the functions, O&M capabilities, and enhanced performance and stability the latest version provides. If you have any questions, submit a service ticket.

Upgrade Instructions

https://support.huaweicloud.com/intl/en-us/redisug-nosql/nosql_02_0732.html

Estimated Interruption

A single node is briefly interrupted (no more than 10s). The actual duration and number of interruptions depend on how many instance nodes you have.

Estimated Upgrade Duration

The upgrade takes about 30 min. The actual duration depends on how many instance nodes you have.

Upgrade Impacts

1. Upgrading the minor version of an instance will restart each node of the instance in sequence. When a node is being restarted, its services will be taken over by another node. Each takeover will interrupt services (for less than 10s). So, perform an upgrade during off-peak hours and enable automatic reconnection so that each node can be reconnected immediately after being restarted.

2. For details about the upgrade operations and impacts, visit https://support.huaweicloud.com/intl/en-us/redisug-nosql/nosql_02_0732.html