Questions regarding system operation design are explained.
Q: Is operation possible without using GUI (GDS Management View)?
Q: Is it possible to perform GDS Snapshot operations through the GUI (GDS Management View)?
Q: Do I need to specify certain settings in order to perform a hot swap?
Q: Are the mirrored disks differentiated as "primary" and "secondary" disks?
Q: I want to know GDS resident daemons to monitor resident processes.
Q: Is operation possible without using GUI (GDS Management View)?
A
Yes. Two types of operation management interfaces are available: the GUI (Graphical User Interface) and the CLI (Command Line Interface).
Q: Is it possible to perform GDS Snapshot operations through the GUI (GDS Management View)?
A
Proxy object operations are possible. To operate shadow objects, use the CLI (Command Line Interface).
Q: Can disks with different performance specifications, such as size and revolution speed, be mirrored?
A
Yes. However, when mirroring disks of different sizes, you will only be able to use the capacity of the smaller disk. For example, if a 4 GB disk and a 9 GB disk are mirrored, only 4 GB out of the 9 GB disk will be accessible.
Also, when mirroring disks with different specifications such as revolution speed, the read performance becomes unbalanced and the write performance will depend on the slower disk's performance. Therefore, it is recommended to mirror disks with the same specifications.
See "A.3.1 Guidelines for Mirroring" also.
Q: Do disks have to have equivalent attributes, such as size and revolution speed when I relate them as master and proxy?
A
You can relate disks as master and proxy even if they have different attributes, such as size and revolution speed.
However, it will result in that the read performance becomes unequal and the write performance depends on the slower disk's performance. Therefore, it is recommended to relate disks with the same attributes wherever possible.
For details see also "A.1.8 Proxy Configuration Preconditions."
Q: Can I combine mirroring and striping?
A
You can mirror stripe groups.
However, you cannot stripe across multiple mirror groups, using them as stripe columns.
Q: Do I need to specify certain settings in order to perform a hot swap?
A
No.
As long as the disk you are using supports hot swap, no special settings are required.
For procedures, see "5.3.4 Disk Swap" and "D.8 sdxswap - Swap disk."
Q: What is the purpose of single volume?
A
GDS is software which improves availability and manageability of the disk data. Single volume is used when users need to improve manageability, rather than availability. For details, see "1.3.2 Centralized Disk Management."
For example, if availability is already realized by using disk arrays, the user will benefit by using the physical disk as a single volume excluded from mirroring, to improve its manageability.
Using GDS to manage the volume as a single volume not only ensures consistent management, but also allows you to include the volume in the mirroring configuration without stopping the service application, when the need arises.
Q: Can the messages output by GDS driver, daemon and command be displayed in a language other than English?
A
No. The messages are in English only.
Q: Are there any restrictions on the LUN (Logical Unit) configuration within the disk array when managing the disk array using GDS?
A
There are no restrictions on LUN RAID level or size.
Q: Are the mirrored disks differentiated as "primary" and "secondary" disks?
A
No. Disks mirrored with GDS are handled in the same way. Both the user and the application do not differentiate between the two.
However, the only exception is the system disk. The system disks may be differentiated as the primary and the secondary disks to initially be booted with OpenBoot. For details, see "F.1.5 System Disk Abnormality."
Q: When reading the mirror volume, which one of the slices in the volume will the I/O request be issued to?
A
When a read is issued to a volume, the last access block number of each disk in the volume will be checked. The slice with the value closest to the read block number will be read. In other words, disk with the shortest seek distance will be read first.
When a write is issued to a volume, it will write to all slices in the volume (except from slice in ACTIVE status) and will return the write results after completing the write request on all slices.
Q: When mirroring a disk array that supports the hot spare function, which is preferable; using the spare disk of the disk array, or that of GDS?
A
Generally, using the spare disk of the disk array is recommended.
Q: Which of the GDS messages should be monitored?
A
Generally, the following messages given in the "GDS Messages" should be monitored.
PANIC or WARNING level messages and Internal error messages shown in "E.2 Driver Messages"
HALT, ERROR, or WARNING level messages and Internal error messages shown in "E.3 Daemon Messages"
For details, see "Appendix E GDS Messages."
Q: I want to know GDS resident daemons to monitor resident processes.
A
The following table contains GDS resident daemons and the description.
Name | ps -ef command's CMD | Description |
|---|---|---|
sdxmond | /usr/sbin/sdxmond | Monitors GDS daemons |
sdxservd | /usr/sbin/sdxservd | Required daemon |
sdxlogd | /usr/sbin/sdxlogd | Required daemon |
sdxexd | /usr/sbin/sdxexd | Required daemon |
sdxclc | /usr/sbin/sdxclc -W | Cooperates with PRIMECLUSTER CF |
sdxcld | /usr/sbin/sdxcld -W | Cooperates with PRIMECLUSTER CF |
sdxcle | /usr/sbin/sdxcle -f n | Cooperates with PRIMECLUSTER CF |
sdxhealthcheckd | /etc/opt/FJSVsdx/bin/sdxhealthcheckd | Daemon monitoring the GDS daemon and driver hang-ups. |
GDS: Global Disk Services
If any resident daemon other than sdxmond and sdxhealthcheckd terminate abnormally, sdxmond automatically restarts the daemon and outputs a message as follows to the GDS log file /var/opt/FJSVsdx/msglog/sdxservd.log.
When succeeded in restart
SDX:sdxmond: WARNING: respawned daemon daemon successfully |
When failed in restart
SDX:sdxmond: HALT: failed to respawn daemon daemon, osfunc=osfunc, errno=errno |
If sdxmond terminates abnormally, the OS init(1M) process automatically restarts the daemon. In that event, no message is output.
If sdxhealthcheckd terminates abnormally, the OS svc.startd(1M) process automatically restarts the daemon. In that event, no message is output.
As described above, the GDS resident daemons are automatically restarted in the event of abnormal termination, and it is not necessary to monitor them.
In addition, GDS Snapshot has no resident daemon.
See
For the meanings of the messages and the resolutions, see "E.3 Daemon Messages."
Q: When configuring a 24-hour operation system, would files of incremental size such as message log file have any effect?
A
Since GDS's message log file is managed so that it will not exceed a certain size, there are no effects that need to be considered.
Q: Can NAS be mirrored?
A
NAS cannot be mirrored.
The disk device managed by GDS is the one that can be seen from the server where GDS operates just like the internal disk.
NAS such as ETERNUS NR1000F series is not supported because it can be seen as a file server from the server where GDS operates.
