DL380G6 / Centos5.5: still surface scanning after 2 months?

[Harald Krake]

As far as I remember from older proliants, the smartarray surface scan may take a few days.
However, our new DL380 G6 still keeps on scanning after two months! Furthermore, changing the surface scan delay via hpacucli seems to have no effect at all.

The config:
DL380 G6 with latest firmware
4 x 146G Raid5 + 1 hotspare
onboard P410i with 512MB FWBC
Centos 5.5 (x86_64) with the standard cciss-driver (3.6.20-RH4)

I already tried to install the latest cpq_cciss-3.6.24-5.rhel5.x86_64.rpm
but it refuses to install because the kernel (2.6.18-194.11.1.el5) isn't supported yet.

Any help appreciated...

Harald

[Jimmy Vance]

Anytime the drives are idle for x amount of time, the Smart Array will start a surface scan.

[Harald Krake]

hmm... the two ML530 that were replaced by the DL380 two months ago definitely did the scanning only for the first couple of days after the logical drive had been created
(and perhaps after a certain amount of time).
Furthermore, you could see the surface scan delay working, for example 3 seconds pause before the scan continued.

In contrast, the green drive-activity leds on the DL380 are constantly on, except when there is some I/O-activity. There is _no_ visible surface scan delay, regardless what has been configured in the controller settings (0s, 3s, 15s, whatever, makes no difference).

Was that behaviour changed with newer SmartArray-controllers?

What is the "surface scan delay" parameter still good for, if it's ignored?

[Jimmy Vance]

The function has not changed that I'm aware of.

From the Smart Array configuration guide.

"The setting for the surface scan delay determines the time interval for which a controller must be inactive
before a surface scan analysis is started on the physical drives that are connected to it.
Surface scan analysis is an automatic background process that ensures that you can recover data if a
drive failure occurs. The scanning process checks physical drives in fault-tolerant logical drives for bad
sectors, and in RAID 5 or RAID 6 (ADG) configurations, it also verifies the consistency of parity data."

[Jimmy Vance]

When the surface scan function is active, the drive lights are not steady solid green, they do flash.

[Saberus]

IIRC, a solid green light means that the drive is online and part of an array, nothing more. The green light will flash when the drive is performing an I/O operation.

[Harald Krake]

I checked the docs and that's correct!
Thank you!