Hi, thanks for the suggestion, I read this solution somewhere already, that kinda consists in moving /boot to the internal MMC memory, right?

Does your uBoot find the eSata drive?

I'm wondering if this would work in my case. I did an experiment, I manually loaded my uImage and uInitrd from tftp. That kernel should be able to find SATA drives. It was the kernel that run for the last 200 days. It didn't find the root drive. Before that, my uboot usually found the disk after some probing and I could boot everything.

So I suspect the problem lies elsewhere. Or is there something different in that process that I haven't understood?

Hi,
 
 I have an eSata Sheevaplug that I use as a server for a year, now.
 
 I always had some trouble booting with the eSata, I had to start a couple of times by powering up the external disk and then the sheevaplug 'till uboot found the disk and booted.
 
 I'm not an expert, but it feels like some timing problem where the Sheevaplug can't find the disk when it spun down and can't start it to spin again.
 
 I was able to start the Sheevaplug like this and it ranfor the last 227 days. Unfortunately, I had to move some furniture around this weekend, so I had to bork my nice sheevaplug uptime by shutting it down, unplugging it, move it around to it's new location, plug it in again and boot...
 
 ... and nothing! As much as I tried, I wasn't able to boot again from the eSata connection. I tried with the newit uboot (I think it's the marvell version) and updated with the one from upstream, as advised here:
 http://www.cyrius.com/debian/kirkwood/sheevaplug/
 ( that's the tutorial I followed to install my plug).
 
 From the uboot command line, "ide reset" never found the drive, but I had two different cases, behaviour being constant until next reset of the plug:
 "ide info" finds no drive
 "ide info" finds the drive, displaying it's ID and stuff, but when trying to read from it something (e.g. "ide part"), it doesn't work, with errors like "Device not ready". In this case,
 
 That drive is a WD Caviar Green, which has some specific firmware that might make it spin down more often. From another machine, I tried to change the timeout for the spin down with hdparm, but it didn't help either.

As I use it for my mailserver, I needed it to be running, so it's now running from the USB, as my HDD case also supports this, but has slower access time.

Has anyone had similar problem or has an idea on what I could do to get my eSata back again?

Thanks,

nab

I ordered my SheevaPlug mid-november 2010 from NewIT, so from what I read on the NewIT forum, it should have the new PSU. Haven't found an easy way to check that, thought...

So, with my external HD, I shouldn't expect it to die anytimes... No big need to order a replacement PSU or replacement sheeva just in case.

Thanks for your answer!

nab

Hi,

I have an eSata SheevaPlug running as a server for a month and a half now from a debian squeeze installation. Its running from an external eSata harddrive with its own powersupply. I read that some people had dead sheevaplug here.

I was wondering if people had SheevaPlugs running 24/7 here and for how long. I have an old machine that I can use as an emergency server, but was wondering what the likeliness of it happening was.

Thanks for your info,

nab