ACOS 6.3.0

Arcfra Cloud Operating System>
CLI command>
Managing physical disks

Managing cache partitions

In SSDs, in addition to journal partitions that record operations, some partition spaces are used as cache to accelerate data reads and writes.

Viewing node cache information

View the sizes and proportions of active cache, inactive cache, clean cache, and sufficient cache on a node.

Note:

The sum of the above four metrics may differ from the total cache size because a portion of the cache partition is reserved as the cluster's performance partition, which is excluded from the first three metrics.

Procedure

Run the following command on a cluster node:

zbs-chunk [--ins_id <id>] cache list

Parameter	Description
`--ins_id <id>`	Optional. The physical disk pool ID. This parameter can be used to filter and display information for a specific physical disk pool in environments with multiple physical disk pools.

Output example

====ALL INSTANCES CACHE SUMMARY====
CACHE HIT RATE:        1.0
TOTAL ACTIVE:           0B ( 0.00%)
TOTAL INACTIVE:         0B ( 0.00%)
TOTAL CLEAN:            0B ( 0.00%)
TOTAL PERF USED:   681.07G ( 6.04%)
TOTAL FREE:         10.34T (93.96%)
TOTAL:              11.01T
========  ==============  =================================================  ====================================  ===============  =============  ==========  =============  ====================================  ===========  ============  ===========
  INS ID  PATH            DEVICE ID                                          UUID                                    NUM IO ERRORS  STATUS           ERRFLAGS    NUM SLOW IO  PART UUID                               WARNFLAGS  TOTAL SIZE    USED SIZE
========  ==============  =================================================  ====================================  ===============  =============  ==========  =============  ====================================  ===========  ============  ===========
       1  /dev/nvme0n1p4  nvme-eui.01000000000000005cd2e4fbb1e35551-part4    47cc2154-ca41-456f-80ab-cb1b53916e4f                0  CACHE_MOUNTED           0              0  6a1c9e11-1964-456c-89da-8d6f6d934ffb            0  2.61 TiB      321.07 GiB
       2  /dev/nvme1n1p4  nvme-INTEL_SSDPF2KE032T1_BTAX3161016C3P8CGN-part4  e9cc8c41-c9ef-4a65-9773-82ba7f8ef405                0  CACHE_MOUNTED           0              0  ced654e4-6b2d-4344-8627-6a8e57b41e44            0  2.61 TiB      120.00 GiB
       3  /dev/nvme2n1p2  nvme-INTEL_SSDPF2KE032T1_PHAX331404UY3P8CGN-part2  d4f180ad-5b8d-40af-9568-ec870db4716e                0  CACHE_MOUNTED           0              0  9fb6b81a-8e6d-4d85-94de-b4a628d1a6eb            0  2.89 TiB      120.00 GiB
       4  /dev/nvme3n1p2  nvme-INTEL_SSDPF2KE032T1_PHAX325004L53P8CGN-part2  e3743e81-fab4-42c2-b8e2-ecf493ae446c                0  CACHE_MOUNTED           0              0  644079ef-5078-4c89-a14e-dadd9a3d4cd7            0  2.89 TiB      120.00 GiB
========  ==============  =================================================  ====================================  ===============  =============  ==========  =============  ====================================  ===========  ============  ===========

Output note

Parameter	Description
`CACHE HIT RATE`	The cache hit rate. The cache represents read cache only.
`TOTAL ACTIVE`	The active cache size and its percentage (excluding performance partitions). Expected to remain 0 in the long term.
`TOTAL INACTIVE`	The inactive cache size and its percentage (excluding performance partitions). Expected to remain 0 in the long term.
`TOTAL CLEAN`	The clean cache size and its percentage (excluding performance partitions). Expected to be greater than 0 in the long term.
`TOTAL PERF USED`	The total space used by the performance tier.
`TOTAL FREE`	The sufficient cache size and its percentage (including performance partitions).
`TOTAL`	The total cache size (including performance partitions).
`PATH`	The partition path.
`DEVICE ID`	The physical disk ID.
`UUID`	The unique UUID identifier of the cache partition.
`NUM IO ERRORS`	The number of I/O errors.
`STATUS`	The partition status.
`ERRFLAGS`	The error flags.
`NUM SLOW IO`	The number of slow I/Os.
`PART UUID`	The unique UUID identifier of the physical partition.
`WARNFLAGS`	The alarm flags.
`TOTAL SIZE`	The total capacity of the cache partition.
`USED SIZE`	The used space.

Formatting the cache partition under the specified path

Procedure

Formatting the cache partition under the specified path

Run the following command on a cluster node to format the cache partition under the specified path:

zbs-chunk [--ins_id <id>] cache format <path>
Format the cache partition with existing data under the specified path

Run the following command on a cluster node to format the cache partition with existing data under the specified path:

zbs-chunk [--ins_id <id>] cache format --force <path>

Note:

Running this command will erase all existing data on the cache partition. Proceed with caution.

Information:

In environments with multiple physical disk pools, if the specified partition is mounted by another physical disk pool, the command will fail. In this case, you must unmount the corresponding partition before running the command.

Parameter	Description
`--ins_id <id>`	Optional (required in a multi-physical disk pool environment). The physical disk pool ID. This parameter specifies the physical disk pool used for formatting in a multi-physical disk pool environment.

Output example

Cache successfully formatted

Mounting a cache partition to a chunk

Procedure

Run the following command on a cluster node to mount a cache partition to the chunk under the specified path:

zbs-chunk [--ins_id <id>] cache mount <path>

Parameter	Description
`--ins_id <id>`	Optional (required in a multi-physical disk pool environment). The physical disk pool ID. This parameter specifies the physical disk pool used for mounting in a multi-physical disk pool environment. It must be consistent with the physical disk pool used for formatting.

Output example

Successfully mounted

Unmounting a healthy cache partition from a chunk

Procedure

Run the following command on a cluster node to unmount the cache partition under the specified path from the chunk:

zbs-chunk cache umount <path>

Output example

Cache device is STAGING now, umount will be done in background

Stopping the unmount of a healthy cache partition from the chunk

Procedure

Run the following command on a cluster node to stop unmounting the cache partition under the specified path from the chunk:

zbs-chunk cache cancel-umount <path>

Output example

Cache cancel umount submitted

Unmounting an abnormal cache partition from a chunk

Procedure

Run the following command on a cluster node to unmount the cache partition under the specified path from the chunk. This operation cannot be undone:

zbs-chunk cache umount --scheme uuid --mode offline <uuid>

Output example

Cache device is STAGING now, umount will be done in background

Setting the cache partition to a healthy status

Procedure

Run the following command on a cluster node to set the cache partition under the specified path to a healthy status:

zbs-chunk cache set-healthy <path>

Output example

Cache device is set healthy!

Setting the cache partition to an unhealthy status

Procedure

Run the following command on a cluster node to set the cache partition under the specified path to an unhealthy status:

zbs-chunk cache set-unhealthy <path>

Output example

Cache device is set unhealthy! umount and recover will be done in background

In this article

Viewing node cache information
Formatting the cache partition under the specified path
Mounting a cache partition to a chunk
Unmounting a healthy cache partition from a chunk
Stopping the unmount of a healthy cache partition from the chunk
Unmounting an abnormal cache partition from a chunk
Setting the cache partition to a healthy status
Setting the cache partition to an unhealthy status