이 기사가 도움이 되었습니까?
이 기사를 더욱 유용하게 만들 수 있는 방법은 무엇입니까?
You can use lmacli to configure a SAS or FC data path between a Linux host and a Lyve Mobile Rackmount Receiver connected to the same data network.
The following steps are required:
To install multipath tools:
sudo yum install lsscsi device-mapper-multipath
To install multipath tools:
sudo apt install lsscsi multipath-tools
The connection type for your Lyve Mobile Array is displayed as SAS or FC.
To view device details:
lmacli info IP address
where IP address is a Rackmount Receiver port IP address, for example:
lmacli info 192.168.0.100
Example output | |
---|---|
=== Identification === | |
Product | : STJX46000400 |
Serial number | : NB261234 |
Connection type | : SAS |
Name | : |
=== Disk encryption === | |
Status | : Secured, unlocked |
=== Pools === | |
Name | : dg |
Total size | : 59.9TB |
Available size | : 0B |
Number of volumes | : 1 |
=== Volumes === | |
Name | : volume-0 |
Pool name | : dg |
Size | : 59.9TB |
RAID level | : RAID0 |
Status | : OK |
Mapped | : No |
To apply the configuration:
lmacli rackmount-setup IP address
where IP address is a Rackmount Receiver port IP address, for example:
lmacli rackmount-setup 192.168.0.100
Example output |
---|
Checking device configuration (SAS) Clearing existing rackmount configuration Creating new rackmount configuration |
To view connected storage devices:
lsscsi
Example output | |||||
---|---|---|---|---|---|
[0:0:13:0] | enclosu | SEAGATE | STJX46000400 | M100 | - |
[0:0:14:0] | enclosu | SEAGATE | STJX46000400 | M100 | - |
If the host is equipped with a dual-port SAS or FC host bus adapter, Linux should display two enclosures. Each enclosure may contain a disk if the cable connects the host to the corresponding ports on Rackmount Receiver. The enclosures and disks associated with the Lyve Mobile Array volume have their vendor attribute set to SEAGATE and their model starts with STJX.
If enclosures are displayed but no disk is associated with them, the host storage configuration must be rescanned in order to refresh the configuration.
To refresh the configuration:
rescan-scsi-bus.sh
Example output |
---|
[SNIP] 2 new or changed device(s) found. [0:0:13:1] [0:0:14:1] 0 remapped or resized device(s) found. 0 device(s) removed. |
Following the rescan, a summary of the storage configuration changes is displayed. In the example above, two new disks are detected (one for each enclosure).
Enter lsscsi again to display the block device associated with each disk:
Example output | |||||
---|---|---|---|---|---|
[0:0:13:0] | enclosu | SEAGATE | STJX46000400 | M100 | - |
[0:0:13:1] | disk | SEAGATE | STJX46000400 | M100 | /dev/sda |
[0:0:14:0] | enclosu | SEAGATE | STJX46000400 | M100 | - |
[0:0:14:1] | disk | SEAGATE | STJX46000400 | M100 | /dev/sdb |
Each disk listed in the example above represents a distinct path to the same LUN exported by the Lyve Mobile Array.
The multiple paths to the device are coalesced by the device mapper multipath daemon. See Configure I/O multipath to improve I/O bandwidth below.
The device mapper multipath daemon detects and coalesces multiple paths to devices for failover or performance purposes.
To display multipath devices:
sudo multipath -11
Example output |
---|
mpatha (3600c0ff000f392966aae456201000000) dm-1 SEAGATE,STJX60000400 size=55T features='0' hwhandler='1 alua' wp=rw |-+- policy='service-time 0' prio=50 status=active | `- 0:0:13:1 sda 8:0 active ready running `-+- policy='service-time 0' prio=50 status=enabled `- 0:0:14:1 sdb 8:16 active ready running |
To improve I/O bandwidth, it is recommended to change the path_grouping_policy parameter from the default failover to group_by_prio. To set the policy to group_by_prio for all Lyve Mobile Arrays, add the following snippet to /etc/multipath.conf:
Example output |
---|
devices {
|
After restarting the device mapper multipath daemon, the result of the multipath -11 command should look like this:
Example output |
---|
mpatha (3600c0ff000f392966aae456201000000) dm-1 SEAGATE,STJX60000400 size=55T features='0' hwhandler='1 alua' wp=rw `-+- policy='service-time 0' prio=50 status=active |- 0:0:13:1 sda 8:0 active ready running `- 0:0:14:1 sdb 8:16 active ready running |
The first line starts with the multipath device name (mpatha in the example above). The full path to the device is /dev/mapper/mpatha.