Modifying Linux SCSI 'spinning up disk' delay

[tlemons]

Hi

I’m using a Red Hat Linux 7.3-based system with a CLARiiON CX300 which has 8 LUNs. /dev/sd* devices have been created for each path to each LUN. Since only one SP serves a LUN at a time, all of the paths to a LUN through ‘the other’ SP are not available. During system boot, whenever Linux attempts to access these not-available paths, I see the following:



Jun 22 09:15:49 WasabiD kernel: sdb: Spinning up disk...........................

............................................................................not responding...

Jun 22 09:15:49 WasabiD kernel: SCSI device sdb: 1052219904 512-byte hdwr sectors (538737 MB)

Jun 22 09:15:49 WasabiD kernel: sdb:<6>Device 08:10 not ready.

Jun 22 09:15:49 WasabiD kernel: I/O error: dev 08:10, sector 0

Jun 22 09:15:49 WasabiD kernel: Device 08:10 not ready.

Jun 22 09:15:49 WasabiD kernel: I/O error: dev 08:10, sector 0

Jun 22 09:15:49 WasabiD kernel: unable to read partition table





Problem is, all of those dots after ‘Spinning up disk’ take a total of 1.6 minutes per LUN to execute. Do you know of a way to shorten this time period? A co-worker thought this timeout value might be in the SCSI driver. Thoughts on that?



Thanks!

tl

 

[unixfreak]

Why not remove the 'wait spin' on the drives instead?

Cheers

 

[tlemons]

Thanks for the suggestion. Sorry to be a dim bulb, but how would I do that?

Thanks
tl

 

[unixfreak]

There's usually a jumper on the bottom of the drive (or rear). Somewhere on the drive it should say which pin is the spin delay or wait spin.

Cheers

 

[tlemons]

Ah, thanks. That's not viable in this case, as the EMC CX300 is a storage system, not an individual disk.

Thanks
tl