Disconnecting eSata drive

[T-Rhex]

Hello, I'm testing out HDS on my system and so far only one issue has come up for me. I have external eSata drives that I turn on/off and swap in/out on a regular basis. I'm using WinXP (which doesn't officially support this) and found a way to hot swap the drives: turning on isn't a problem, and before turning the drive off I uninstall it using Device Manager. This has been working perfectly for years.

However, while HDS is running, as soon as I uninstall the drive, it is redetected as new hardware and becomes active again. I don't want to risk powering it off (it never turned itself off before) without uninstalling it first, but to do this I have to shut down HDS first.

Would there be any setting that might help? Or is this behavior unavoidable?

WinXP SP3
HDS v3.70 (4981) Pro trial
Intel(R) ICH8R/ICH9R/ICH10R SATA RAID Controller

Thanks,
tr

[hdsentinel]

Hi,

Thanks for your message.

This is interesting as at this computer I have 2 x eSATA hard disks and experience no such problems.
However, these are connected to _real_ eSATA connectors, controlled by Marvell 61xx and JMicron JMB36x controllers, which provide the "safely remove device" function (tray icon) I can use.

The problem is that Intel controller does not really like hot swap/hot plug functionality. According our tests, the device removal / add new hardware method works - but then the newly inserted device performs extremely slow (10 Mbyte/sec or below). I'm not sure if you experienced this or similar ?

But we never experienced that the removed drive (when using this method to remove the eSata drive from the Device Manager) is automatically detected again.

I suspect this may be a driver issue for the Intel controller. Not sure about the current driver, but you may consider to update to the latest driver from the webpage of Hard Disk Sentinel, if you select Support -> Driver Zone.

Or, an even better method would be to use a real eSATA controller, like an above mentioned JMB36x controller with eSATA port(s). Personally I'd recommend that instead of using Intel controller for hot swap/hot plug.

Anyway, if you can use Report menu -> Send test report to developer option, I'd happily check the status and the current version of the driver to test and reproduce the situation.

[T-Rhex]

Thanks for your quick response, even on a Sunday!

I've never really paid attention to speed, as I usually turn on the external drive, start the backup, and walk away. That was before I tried HDS, so now I can actually monitor the transfer rate. My MB doesn't have an alternate Sata controller; can I freely add on a 2-port eSata PCIe card without any drawbacks? Are those controllers you mentioned preferred? Some available locally are Silicone Image, and a card from Bytecc which supports Sata III doesn't list the chipset. There is a 1-port Startech card using the JMB363.

I've sent the test report.

Thanks very much!
tr

[hdsentinel]

Thanks, I'm glad if I could help

I examined the report and yes, I saw that the Intel controller has a bit older driver (version 8.xx, released in 2008).
The latest Intel driver (10.xx) from the Driver Zone ( http://www.hdsentinel.com/driverzone.php or directly from http://www.hdsentinel.com/driver/stor_a ... 026_pc.zip ) may work differently.

Yes, I can confirm that if you have free PCI / PCIe slot, you can freely add a such controller without any drawbacks. That will provide eSATA port(s) you can use, just install the drivers for that (also from the Driver Zone )
The JMB363 controller would be fine - this chip is also used on many motherboards for integrated eSATA solution. The external PCIe card should also work without problems.

[T-Rhex]

Thanks for the feedback, much appreciated.

I'll look into the drivers you mentioned. I did notice, however, that connecting/disconnecting/reconnecting drives the way I've been doing it has no impact on transfer rates; I was getting 30-40MB/s doing some extensive file compares before and after reconnecting the drive.

One feature request came to mind: the average transfer rate is a good reading during transfers. However, when the transfers stop, that data is lost (ie. the average is total/time). Would you consider adding an actual average transfer rate which only takes non-zero rates into account? ie. an average of non-zero rate readings, as opposed to ( total transfer ) / ( elapsed time ). Or is it there somewhere already?

Thanks again.
tr

[hdsentinel]

Thanks for the suggestion about calculating and displaying the actual average values, it will be considered and may be added in a later version.