I had more massive troubles yesterday: I decided to synchronized the array Friday night. Since it is a long operation, I decided to let it run overnight. Every day at 01h00 I do a back up on tape of my most crucial data. On Saturday morning when I checked the back up, Back Up Exec indicated that part of the back up failed. I checked the directory whose back up failed and I was appalled to see that the data on my array (where these directories were) was unreadable. I rebooted to force a chkdsk. Upon rebooting, the chkdsk was not done. I started the Disk Management tool to check if the disk was OK and then saw that the disk was unreadable. I tried everything I could, peered into XP documentation and found nothing of any help. I sought advises from a couple of friends and none could tell me how to salvage my array. I decided to reformat the array (again) and forgo the data there. I didn't lost too much (I could get most of the data back from tapes) but still it is way too much. What really pisses me off in this whole adventure is the incredible amount of time I wasted. I always believed that when you buy high quality parts, it is easier than with el cheapo parts. The problem here is not quality but inter-operability: It is not because you get high quality, because you do all the necessary preliminary research that everything will go without problem. Most of the stuff you read on the 'net if for average John Doe. If you are somebody with a particular high-end need (digital video editing in my case) the information you need is probably not available. How many people run on the same dual CPU motherboard an SX6000 and a DV500PLUS? To my knowledge I am the only one. How many people run on a single CPU motherboard an HPT370 controller and an "all in one" video card? Probably a lot... I learned by myself:
With 7 HDs, it is better to have two average power supply than one good one.
The Promise SX6000 doesn't play well with Phoenix BIOSes.
The Promise SX6000 doesn't work at all with version 4.01 of the said BIOS on the Tyan S2466M-4N. You must upgrade to 4.03
At advogato.org, sarum wonders why there is no Simple SQL Protocol. What sarum wants is a simple way to pass SQL statement and dataset around in an implementation-language independant way. A way to do it would be to encapsulate the access to the database in an XML-RPC or SOAP server. A SQL statement, a database name, a username/password could be passed to a server who would "convert" this into a call using the database protocol native to the server and pass the result back to the client as a string with known delimiters between fields and records. It could allow any SOAP or XML-RPC capable language to fetch data from an M$ SQLServer whose capabilities would have been encapsulated and exposed by an XML-RPC or SOAP server. It could also let VS.NET access MySQL or Postgres running under Linux. I'll give it a try.
