[rescue] A tale of two Sun A5000's

[Mr Ian Primus]

Well, when dealing with a fault you can't find, take it apart and put it back together...

I stripped the faulty array down to the chassis last night - removing everything, including the backplanes in order to get a look at the interconnect board. After removing EVERYTHING, you can unbolt the horizontal interconnect and lift it out, still attached to the vertical interconnect.

The only oddity that I could find was on the vertical interconnect board. Along the bottom of the board are four large orange surface mounted capacitors. The bottoms of all four of these are soldered to a thin bit of metal shielding that's held in place with a screw, and bent down such that it would make contact with the bottom of the chassis. All four of these solder points were broken, and the shielding was only held in with it's screw. I resoldered the thing to the best of my ability with the tools at my disposal (it's really hard to solder because of the heat-sinking properties of the metal). 

Also while I was in there I cleaned all the dust out of the chassis, cleaned the crud out of the disk guides, and blew out all the connectors with canned air.

Reassembled the whole thing, hook it all back up and put disks back in it. It all seems to be working. It's loaded with a mix of drives, mostly 9's I had laying around, and I've configured it as one big RAID-0 array. I've been writing lots of data to it to see if I get any more I/O errors. So far, so good.

I can't believe that such errors could actually have been caused by those cracked solder points on the ground shield - more likely the reseating of connectors was more at play here. It's very possible this particular disk array was dropped at some point after becoming obsolete.

Interesting (and obvious, to anyone who's seen an A5000) to note is that the empty metal frame of the array, even with no electronics, is still suprisingly heavy. Certainally a well made bit of metalwork.

-Ian