Modify unit configs

Modifying unit configs means increasing or decreasing the CPU, memory, log disk capacity, etc. for unit configs.

Considerations

Before modifying a unit config, if the unit corresponding to the unit config is being used by a tenant and additional resources are needed, you must ensure that there is sufficient remaining resources available for allocation on each node during the resource increase process. You can query the total resources and allocated resources on each node through the oceanbase.GV$OB_SERVERS view, and then determine whether it is possible to modify the unit config.

If you need to increase a unit config, you must ensure that the total amount of resources after the increase meets the following requirements:

Sum(min_cpu) <= CPU_CAPACITY;
Sum(max_cpu) <= CPU_CAPACITY * resource_hard_limit;
Sum(memory_size) <= MEM_CAPACITY;
Sum(log_disk_size) <= LOG_DISK_CAPACITY;

where

• CPU_CAPACITY: Indicates the total CPU capacity on a single node.

• resource_hard_limit: Indicates the amount of CPU resources allocated by the system when allocating units. The value range is [100, 10000]. The default value is 100, indicating that over-allocation is not allowed.

  For more information about the resource_hard_limit parameter, see resource_hard_limit.

• MEM_CAPACITY: Indicates the total memory capacity on a single node.

• LOG_DISK_CAPACITY: Indicates the total log disk capacity on a single node.

  The total log disk capacity is jointly controlled by the cluster-level parameters log_disk_size and log_disk_percentage. log_disk_size is used to set the size of the disk where redo logs are stored, with a default value of 0. log_disk_percentage is used to set the percentage of the disk space occupied by redo logs, with a default value of 0. The usage rules for these two parameters are as follows:

  • If the value of log_disk_size is 0M and that of log_disk_percentage is not 0, the system allocates the log disk space based on the value of log_disk_percentage.

  • If the value of log_disk_size is not 0M, the system allocates the log disk space based on the value of log_disk_size, regardless of whether the value of log_disk_percentage is 0.

  • If the value of log_disk_size is 0M and that of log_disk_percentage is 0, the system automatically calculates the percentage of the disk space occupied by redo logs based on whether the log and data share the same disk.

    • If they share the same disk, the percentage of the disk space occupied by redo logs is 30%.

    • If redo logs exclusively occupy the disk, the percentage of the disk space occupied by redo logs is 90%.

  For more information about the log_disk_size parameter, see log_disk_size.

  For more information about the log_disk_percentage parameter, see log_disk_percentage.

If you need to decrease a unit config, you must ensure that the total amount of CPU, memory, log disk capacity after the decrease is greater than or equal to the currently used log disk space.

Limitations

Only the sys tenant can modify unit configs.

Procedure

When modifying unit configs, you can modify several items in CPU, memory, IOPS, and log disk capacity. The items that are not modified will remain unchanged.

1. Log in to the sys tenant of the cluster using the root user.

2. Execute the following statement to modify a unit config:

   ALTER RESOURCE UNIT unitname
   MAX_CPU [=] cpunum,
   [MIN_CPU [=] cpunum,]
   MEMORY_SIZE [=] memsize,
   [MAX_IOPS [=] iopsnum, MIN_IOPS [=] iopsnum,IOPS_WEIGHT [=]iopsweight,]
   [LOG_DISK_SIZE [=] logdisksize];
   

   An example of modifying the configurations of the unit1 resource unit is as follows:

   obclient> ALTER RESOURCE UNIT unit1 MAX_CPU 15, MEMORY_SIZE '20G', LOG_DISK_SIZE '4G';