Hy Peter and all,
after a couple of reboot the PowerPath and reconfiguring the LUN
exported by the Clariion now all seems to working fine.
Following the output of "zpool status":
---
machine# zpool status
pool: tank
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
tank ONLINE 0 0 0
raidz1 ONLINE 0 0 0
c3t5006016941E0222Ed3 ONLINE 0 0 0
c3t5006016141E0222Ed1 ONLINE 0 0 0
c2t5006016841E0222Ed2 ONLINE 0 0 0
c3t5006016141E0222Ed0 ONLINE 0 0 0
errors: No known data errors
machine#
---
Meanwhile, I did some benchmarching on "raidz" zfs pool. Normally,
with application running on a zone where open, read and move thousand
of file (average dimension of that file is around 2k) I gave this
result:
--
machine# zpool iostat 2
capacity operations bandwidth
pool used avail read write read write
---------- ----- ----- ----- ----- ----- -----
tank 7.56G 111G 4 233 25.7K 1.37M
tank 7.56G 111G 0 0 0 254
tank 7.56G 111G 0 866 0 5.35M
tank 7.56G 111G 0 0 0 0
tank 7.56G 111G 0 389 0 1.23M
tank 7.56G 111G 0 0 0 0
tank 7.56G 111G 0 79 0 237K
tank 7.57G 111G 0 769 0 4.29M
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 580 0 1.67M
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 152 0 472K
tank 7.57G 111G 0 631 0 3.36M
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 98 0 439K
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 191 0 505K
tank 7.57G 111G 0 873 0 5.77M
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 163 0 463K
tank 7.57G 111G 0 0 0 0
tank 7.57G 111G 0 252 64.2K 817K
tank 7.57G 111G 0 550 0 4.90M
^C
machine#
---
Can I achieve more performance?
Thanks.
Cesare
On Sun, Jun 22, 2008 at 5:48 PM, Peter Tribble <[EMAIL PROTECTED]> wrote:
> On Sun, Jun 22, 2008 at 2:06 PM, Cesare <[EMAIL PROTECTED]> wrote:
>> Hy,
>>
>> I'm facing to a problem where I configure and create a zpool on my
>> test bed. The hardware is: T-5120 with Solaris10 with latest patch and
>> Clariion CX3 attached by 2 HBA. In this type of configuration every
>> LUN exported by Clariion is viewed 4 times by operating system.
>>
>> If I configure the latest disk by using a controller the "zfs create"
>> doesn't working telling me that there is a devices currently
>> unavailable. If I'll use a different controller (but is the same LUN
>> from the Clariion) I'll not encountered the problem and the raidz pool
>> is created. I'm willing to use that controller for balance the I/O
>> between HBA and storage processor.
>
> My experience is that zfs + powerpath + clariion doesn't work.
>
> (Try a 'zpool export' followed by 'zpool import' - do you get your pool back?)
>
> For this I've had to get rid of powerpath and use mpxio instead.
>
> The problem seems to be that the clariion arrays are active/passive and
> zfs trips up if it tries to use one of the passive links. Using mpxio hides
> this and works fine. And powerpath on the (active/active) DMX-4 seems
> to be OK too.
_______________________________________________
zfs-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
