You are not logged in.

#1 2021-02-12 18:51:54

bazagee
New Member
Registered: 2019-06-17
Posts: 3

RAID performance

History.. I have a workstation at work as a multi VM platform Host. I ran VBox on Windows for some time but decided to change to a light weight Nix Host. The first incarnation was on Ubuntu running hot corners and a few other WYSIWIG's with VBOX but I didn't find it very stable. And I was running software Raid which turned out to be a bad idea.

After some reading and more research I settled on BL Helium and switched to KVM and QEMU for virtualisation. I'll say up front that I'm not well versed as a Linux guy, but I've managed to make things work. I do spend most of my time as an Admin in on my Windows VM in our MSAD environment. Apart from having to restart lightdm after a period of weeks, its a pretty stable platform. Performance however has always been an issue and its only now that I'm really looking into that. I always suspected it was at the hardware RAID level.

I have a SSD running BL with an old LSI Raid card (I think its a M5014 series) from an IBM server with a 3 drive 9TB Raid 5 array used to host guests. A partition with (Ext4) and another partition formatted in NTFS for windows storage.

Natively I'm getting really bad performance from the Array and want to know what to look for and how to go about it in BL? I'm suspecting its some driver issue between BL and the card. When I do a HardInfo lookup on BL I see RAID Bus Controller - LSI Logic/Symbios Logic MegaRAID SAS 2108 [Liberator], so that might be the wrong driver?

SSD:
/dev/sdb:
multcount     = 16 (on)
IO_support    =  1 (32-bit)
readonly      =  0 (off)
readahead     = 256 (on)
geometry      = 21889/255/63, sectors = 351651888, start = 0
Timing cached reads:   27936 MB in  1.99 seconds = 14033.36 MB/sec
Timing buffered disk reads: 746 MB in  3.00 seconds = 248.61 MB/sec

RAID:
/dev/sda:
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0b 00 00 00 00 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
multcount     =  0 (off)
readonly      =  0 (off)
readahead     = 256 (on)
geometry      = 1094187/255/63, sectors = 17578119168, start = 0
Timing cached reads:   24828 MB in  1.99 seconds = 12465.93 MB/sec
Timing buffered disk reads:  62 MB in  4.01 seconds =  15.45 MB/sec

I'm looking up what the sense data error is about now. There's a real good chance I didn't configure the volumes very well from the get go as well.

If anyone has some advice please let me know what other data you would like to see. The Card doesn't have a BBU on it so its not obviously using Write Caching.

Any advice appreciated. Thanks

Offline

Board footer

Powered by FluxBB