better option for storage

[rickygm]

umm , Thanks, I had to make a

You have to create volume groups, not volumes! A filesystem and mounting is not cluster safe and you will loose data! Seriously!

Please follow the guide already provided at:

https://pve.proxmox.com/wiki/Storage_Model#LVM_Groups_-_An_Ideal_Solution

and ommit the iSCSI stuff.

umm , Thanks, I had to make a rollback , and returned to a volume group.
Adamb had told me this before

 

[adamb]

I just want to point out that if you have multiple pipes (which I take as a given, whats the point of multipathing without them?) any one lun can only use one path at a given time, leaving the other(s) idle. For an active/active config you want a minimum of 1 lun per host channel on your storage.

Separate VLAN's and load balancing. Its the recommended approach from my storage vendor. Using multipath also has some other advatages I prefer.

 

[rickygm]

I just want to point out that if you have multiple pipes (which I take as a given, whats the point of multipathing without them?) any one lun can only use one path at a given time, leaving the other(s) idle. For an active/active config you want a minimum of 1 lun per host channel on your storage.

I'll show you what I have in multipath and see if we can improve

[4:0:0:10] disk HP HSV340 1120 /dev/sdj 783GB

[0:0:0:10] disk HP HSV340 1120 /dev/sdb 783GB
[0:0:0:11] disk HP HSV340 1120 /dev/sdc 783GB
[0:0:0:12] disk HP HSV340 1120 /dev/sdd 4.39TB
[0:0:0:13] disk HP HSV340 1120 /dev/sde 4.71TB

[0:0:1:10] disk HP HSV340 1120 /dev/sdf 783GB
[0:0:1:11] disk HP HSV340 1120 /dev/sdg 783GB
[0:0:1:12] disk HP HSV340 1120 /dev/sdh 4.39TB
[0:0:1:13] disk HP HSV340 1120 /dev/sdi 4.71TB

[4:0:0:11] disk HP HSV340 1120 /dev/sdk 783GB
[4:0:0:12] disk HP HSV340 1120 /dev/sdl 4.39TB
[4:0:0:13] disk HP HSV340 1120 /dev/sdm 4.71TB

[4:0:1:10] disk HP HSV340 1120 /dev/sdn 783GB
[4:0:1:11] disk HP HSV340 1120 /dev/sdo 783GB
[4:0:1:12] disk HP HSV340 1120 /dev/sdp 4.39TB
[4:0:1:13] disk HP HSV340 1120 /dev/sdq 4.71TB

multipath -ll
DFAST2 (36001438009b066860000600000690000) dm-2 HP,HSV340
size=730G features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=0 status=active
|- 0:0:0:11 sdc 8:32 active ready running
|- 0:0:1:11 sdg 8:96 active ready running
|- 4:0:0:11 sdk 8:160 active ready running
`- 4:0:1:11 sdo 8:224 active ready running
DFAST1 (36001438009b066860000600000650000) dm-1 HP,HSV340
size=730G features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=0 status=active
|- 0:0:0:10 sdb 8:16 active ready running
|- 0:0:1:10 sdf 8:80 active ready running
|- 4:0:0:10 sdj 8:144 active ready running
`- 4:0:1:10 sdn 8:208 active ready running
DLOW2 (36001438009b066860000600000720000) dm-4 HP,HSV340
size=4.3T features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=0 status=active
|- 0:0:0:13 sde 8:64 active ready running
|- 0:0:1:13 sdi 8:128 active ready running
|- 4:0:0:13 sdm 8:192 active ready running
`- 4:0:1:13 sdq 65:0 active ready running
DLOW1 (36001438009b0668600006000006d0000) dm-3 HP,HSV340
size=4.0T features='1 queue_if_no_path' hwhandler='0' wp=rw
`-+- policy='round-robin 0' prio=0 status=active
|- 0:0:0:12 sdd 8:48 active ready running
|- 0:0:1:12 sdh 8:112 active ready running
|- 4:0:0:12 sdl 8:176 active ready running
`- 4:0:1:12 sdp 8:240 active ready running


I think they have active active.

 

[adamb]

Looks good, I don't see any issues. Just throw LVM on top (Don't create any LV's, just the VG). Then present the VG's to the proxmox hosts from the GUI.

 

[alexskysilk]

I'll show you what I have in multipath and see if we can improve

If I read this right, you have 4 seperate luns each with 4 paths. That seems perfectly balanced. The only POTENTIAL for performance improvements is to split your 4 luns into 8 but that only helps for specific use cases.

 

[rickygm]

sorry, I did not know to explain, I have two hosts, but I have two HBAs per host
lspci | grep -i fibre
07:00.0 Fibre Channel: QLogic Corp. ISP2532-based 8Gb Fibre Channel to PCI Express HBA (rev 02)
0a:00.0 Fibre Channel: QLogic Corp. ISP2532-based 8Gb Fibre Channel to PCI Express HBA (rev 02)

 

[alexskysilk]

rickygm- your config is textbook perfect.

 

[rickygm]

Looks good, I don't see any issues. Just throw LVM on top (Don't create any LV's, just the VG). Then present the VG's to the proxmox hosts from the GUI.

what is your opinion?
http://i1132.photobucket.com/albums/m579/rickygm/look-lvm_zpsmy8dqzba.png

 

[LnxBil]

I would merge the slow and the fast ones in one volume group, but besides that, it's ok.

e.g. like this

root@proxmox4 ~ > pvs
  PV                                       VG       Fmt  Attr PSize PFree
  /dev/mapper/EVA6400_PROXMOX_DATA_FAST_01 san-fast lvm2 a--  1,95t  66,00g
  /dev/mapper/EVA6400_PROXMOX_DATA_FAST_02 san-fast lvm2 a--  1,95t   1,88t
  /dev/mapper/EVA6400_PROXMOX_DATA_SLOW_01 san-slow lvm2 a--  1,95t 991,99g
  /dev/mapper/EVA6400_PROXMOX_DATA_SLOW_02 san-slow lvm2 a--  1,95t   1,20t

 

[rickygm]

OK , thnk to All.