Unsolved
This post is more than 5 years old
2 Intern
•
5.7K Posts
0
553
January 9th, 2008 06:00
How to create new ldevs on a certain type of spindel
I'm dealing with the following:
I need 2TB of mapped storage by tomorrow, made out of 15 metas, each made out of 8 ldevs of 17GB; all ldevs should come from 300GB spindels.
I don't have enough of those 17GB ldevs available, but I do have enough 8.5GB ldevs I don't need anymore. When I check these ldevs, I can see the following:
Device : 2280
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01A:C8 238 44 2908 2 RW N/A 7 286102
05A:C8 238 44 2908 4 RW N/A 7 286102
11A:C16 571 44 2908 1 RW N/A 7 286102
15A:C16 571 44 2908 3 RW N/A 7 286102
}
Device : 22A0
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01A:C8 239 45 2908 1 RW N/A 0 286102
05A:C8 239 45 2908 3 RW N/A 0 286102
11A:C16 581 45 2908 4 RW N/A 0 286102
15A:C16 581 45 2908 2 RW N/A 0 286102
}
I know that the number "286102" is the number of kBytes, so these are the 300GB spindels I want. The number before this 300GB is the disk group number. So in this case I have an ldev in disk group 7 and one in disk group 0.
Another ldev looks like this:
Device : 22A1
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
02D:Ca 296 45 2908 2 RW N/A 7 139828
06D:Ca 248 45 2908 4 RW N/A 7 139828
12D:Ca 296 45 2908 3 RW N/A 7 139828
16D:Ca 296 45 2908 1 RW N/A 7 139828
}
So suddenly my disk group 7 contains 146GB spindels ?
And what about the next one ?
Device : 22A6
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01C:Ca 249 45 2908 1 RW N/A 0 286102
05C:Ca 249 45 2908 3 RW N/A 7 139828
11C:Ca 248 45 2908 4 RW N/A 7 139828
15C:Ca 248 45 2908 2 RW N/A 7 139828
}
Huh ?
I know I created these ldevs myself, but when creating ldevs with a command file you need to specify the disk group number and you cannot mention 2 disk groups, so how did this happen ?
This is the syntax:
create dev
count=64
size=37184
emulation=FBA
config=RAID-5
disk_group=0;
I'm now wondering whether or not the " disk_group" in this command file is the same as the one mentioned in the (partial) "symdev show" output I just mentioned.
How can I be sure that new ldevs will be created on the same type of spindel ?
Any thoughts on this ??
I need 2TB of mapped storage by tomorrow, made out of 15 metas, each made out of 8 ldevs of 17GB; all ldevs should come from 300GB spindels.
I don't have enough of those 17GB ldevs available, but I do have enough 8.5GB ldevs I don't need anymore. When I check these ldevs, I can see the following:
Device : 2280
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01A:C8 238 44 2908 2 RW N/A 7 286102
05A:C8 238 44 2908 4 RW N/A 7 286102
11A:C16 571 44 2908 1 RW N/A 7 286102
15A:C16 571 44 2908 3 RW N/A 7 286102
}
Device : 22A0
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01A:C8 239 45 2908 1 RW N/A 0 286102
05A:C8 239 45 2908 3 RW N/A 0 286102
11A:C16 581 45 2908 4 RW N/A 0 286102
15A:C16 581 45 2908 2 RW N/A 0 286102
}
I know that the number "286102" is the number of kBytes, so these are the 300GB spindels I want. The number before this 300GB is the disk group number. So in this case I have an ldev in disk group 7 and one in disk group 0.
Another ldev looks like this:
Device : 22A1
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
02D:Ca 296 45 2908 2 RW N/A 7 139828
06D:Ca 248 45 2908 4 RW N/A 7 139828
12D:Ca 296 45 2908 3 RW N/A 7 139828
16D:Ca 296 45 2908 1 RW N/A 7 139828
}
So suddenly my disk group 7 contains 146GB spindels ?
And what about the next one ?
Device : 22A6
{
--------------------------------------------------------------
Disk DA Hyper Member Spare Disk
DA :IT Vol# Num Cap(MB) Num Status Status Grp# Cap(MB)
--------------------------------------------------------------
01C:Ca 249 45 2908 1 RW N/A 0 286102
05C:Ca 249 45 2908 3 RW N/A 7 139828
11C:Ca 248 45 2908 4 RW N/A 7 139828
15C:Ca 248 45 2908 2 RW N/A 7 139828
}
Huh ?
I know I created these ldevs myself, but when creating ldevs with a command file you need to specify the disk group number and you cannot mention 2 disk groups, so how did this happen ?
This is the syntax:
create dev
count=64
size=37184
emulation=FBA
config=RAID-5
disk_group=0;
I'm now wondering whether or not the " disk_group" in this command file is the same as the one mentioned in the (partial) "symdev show" output I just mentioned.
How can I be sure that new ldevs will be created on the same type of spindel ?
Any thoughts on this ??
No Events found!
xe2sdc
2 Intern
•
2.8K Posts
0
January 9th, 2008 06:00
hostname:-) # symdisk -sid 123 -by_diskgroup list
The command will list the disks in the backend, grouping them by disk group number.
I find pretty strange that you have group 7 with both 146 and 300 gb disks .. Usually a disk group is made of devices with the same size/speed/interface.
Once identified the disk group from the above output, you can use the "disk_group" option while creating the devices.
Hope it helps ..
-s-
RRR
2 Intern
•
5.7K Posts
0
January 9th, 2008 06:00
A bug in "symdev show" ?
I'm using SymCLI 6.2.1.0. Upgrading to 643 or something will happen as soon as we're upgrading ECC as well.
Message was edited by:
RRR
Added the SymCLI version info
RRR
2 Intern
•
5.7K Posts
0
January 9th, 2008 06:00
I listed all my disk_groups and now I see that each group consists of the same "species". So I wonder what my previous output means.
For example: in my "symdev show" output I can see that spindels 1A:C8 is in disk_group 0 as well as disk_group 7. But when I use the symdisk command you just gave me, I can see spindel 1A:C8 indeed is a 300GB spindel and resides in disk_group 7. Why then does my "symdev show" tell me it's in disk_group 0 for another ldevice ?
xe2sdc
2 Intern
•
2.8K Posts
0
January 9th, 2008 06:00
but I do have enough 8.5GB ldevs I don't need
anymore.
Form a 16way meta .. don't even think about deleting the 8.5 gb devices and creating brand new 17gb devices
xe2sdc
2 Intern
•
2.8K Posts
0
January 9th, 2008 07:00
dynamox
9 Legend
•
20.4K Posts
0
January 9th, 2008 07:00
Form a 16way meta .. don't even think about deleting
the 8.5 gb devices and creating brand new 17gb
devices
why would it be not a great idea ? Your hyper will potentially reside on less number of physical splindles ..but what else ?
RRR
2 Intern
•
5.7K Posts
0
January 9th, 2008 07:00
EMC advised us in this:
it appears that when you have 2 logical devices, a small gap is inbetween each logical device. It's 1 or 2 tracks, but that means, that when you have one 17GB device and delete that one and try to create two 8.5GB ones, you're short a few tracks....
So that's why two times one 8.5GB ldev of ours is a little less than one of our 17GB ldevs. So a meta with 2 members is a bit to small.
We're running a SOD service (Storage on demand) and we need to maintain the sizes we agreed upon.
RRR
2 Intern
•
5.7K Posts
0
January 9th, 2008 07:00
I'll try to locate an ldev with the same weird fenomenon and do a "symdev show" on an already upgraded host as well, just to compare.
I'll let you know.
xe2sdc
2 Intern
•
2.8K Posts
0
January 9th, 2008 07:00
something will happen as soon as we're upgrading ECC
as well.
Right now I can't either confirm or deny if S.E. 6.2.1 have bugs .. But I'm pretty sure that S.E. 6.4.3 is a lot better
RRR
2 Intern
•
5.7K Posts
0
January 9th, 2008 07:00
does anyone know whether or not I can use 6.4.something on an ECC 5.2 SP5 infrastructure host by now ? And if so, which version ?
PS I don't have a separate host I can use for just SE. All systems here are in use by customers.... well almost all of them. Perhaps I need to get my old test server back online just for being able to use SE 643
RRR
2 Intern
•
5.7K Posts
0
January 10th, 2008 06:00
RRR
2 Intern
•
5.7K Posts
0
January 10th, 2008 06:00
Unfortunately I cannot reproduce this output, so I will stop investigating. The symdisk command proved to be helpful ! Thank you, Stefano.