scandisk on Windows 2003

[davis_lambton]

i have a system that is experiencing some serious SQL (2005) db errors pointing to hardware issues. i've been asked to run a scandisk on this system. Is this OK to do in a esx 3.0.1 envinroment?

thx

[esiebert7625 · Accepted solution]

Yes it is. The OS file system has no idea it's on a virtual disk so you should treat it as a physical system.

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

Visit my website:

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

[davis_lambton]

what can be done if the result is 'windows was unable to complete the disk check'?

thanks

ps. other VM's on this ESX host disk check fine.

[esiebert7625]

That could indicate corruption of the OS disk or something running that is interfering with it completing, you can try booting to safe mode and running it again or terminate any running applications when you run it. Otherwise you might try booting of a recovery/boot cd and running it.

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

Visit my website:

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

[davis_lambton]

could this be SQL server preventing a clean diskcheck? i'm a little nervous to start playing with this system with reboots in safe mode, etc, since its been giving me the SQL errors that point to hardware issues. this is a production DB.

i think i'm going to begin buliding another system that can take over for this one in case it goes entirely south on me.

thx.

[esiebert7625]

To be safe I would clone the disk file before doing anything with it so you have a known good backup. You can do this with vmkfstools -i <source disk> <destination disk>, shutdown the VM first before doing this. This is also a good way to try whatever you want without effecting the original VM using a 2nd VM. Follow these steps:

1) Shutdown the SQL VM (VM1)

2) Clone the disk file, vmkfstools -i sqlvm1.vmdk sqlvm1back.vmdk

3) Startup VM1 again

4) Create a new VM (VM2) and tell it to use an existing disk file and browse to sqlvm1back.vmdk

5) On VM2's config change the networking to an internal only vswitch so it is isolated and won't effect your production vm (VM1)

6) Now you can power up VM2 and try whatever you want on it without effecting VM1. When you find what works to fix it you can do it on VM1.

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

Visit my website:

-=-=-=-=-=-=-=-=-=-=-==-=-=-=-=-=-=-=-=-=-=-=-

[davis_lambton]

thanks!

[FredPeterson]

What does "hardware issues" mean? What is leading them to believe disk? Have you done any perfmon counters to see where SQL bottlenecks or disk / memory bottle necks might be occuring?

If you have "bad sectors" in a vmdk, then hrm, wow. Not sure what that means exactly.

[davis_lambton]

two things point to hardware issues, or in this case with VM, virtual disk corruption:

firstly, the SQL error i'm getting is event ID 824 that points to (according to event id description anyway) hardware corruption. either way dbcc checkdb verifies database corruption which i must fix.

and secondly, the failed checkdisk.

this all appears to have begun after a failed scripted backup using vcbMounter. it then left the temp snapshot under snapshot mananger. when i tried to delete this it got to 95% and then eventually failed, crashing the system.