Unsolved
This post is more than 5 years old
3 Posts
0
22108
September 10th, 2008 15:00
Win Server 2003 keep on running Check Disk for Consistency everytimes it restart/power on
Hey guys, am having a problem with the Windows Server 2003 (be it 2003 only or 2003 R2)
Before i write down the issue am encounter, am actually setting up 2 server which is 1 running on Windows Server 2003 and another 1 with Windows Server 2003 R2 (Both are PowerEdge R900), both of them connect to a same Storage Server which is PowerVault MD3000 with 3 set of raid which is hdd ID> 0 & 1 as Raid 1 ; 2 & 3 as Raid 1 ; and 4, 5 & 6 as Raid 5, 14 as Hotspare, and both of the server actually connect to this and using the same drive which is actually sharing all the same setting
Now am having problem like it's keep on running
Check Disk for Consistency /
Disk Consistency Check in between disk G, M, K, L, Q, R, H, J drive; sometimes it check on 1 drive, sometimes it check several drives , it just like randomly check any drive each drive when it reboot or sometimes it just dun check at all(that's what i hope)
Also, all the hdd are brand new and we've checked it doesn't have any bad sector/dirty/improper shutdown
so after several times of testing it, i decided to remove all the virtual disk and it's setting and re-created and mapped it again... but the same problem still come back, it still running the Check Disk for Consistency / Disk Consistency Check intermittently.
and i've try to convert the disk from basic to dynamic aswell, it still not resolve the issue, eventhou that drive been check b4, but after few times of restart it still go and check back that drive again
Any expert know anyway to resolve it?
p/s: sorry for my bad english and description
for those who not sure which disk consistency check i mean, it is the example :
http : / / img127.imageshack.us / my.php?image=diskconsistencyrw1.jpg
Message Edited by yw_sau on 09-10-2008 11:15 AM
Message Edited by yw_sau on 09-10-2008 11:16 AM
Message Edited by yw_sau on 09-10-2008 11:17 AM
Message Edited by yw_sau on 09-10-2008 11:18 AM
Message Edited by yw_sau on 09-10-2008 11:19 AM
No Events found!
Dev Mgr
4 Operator
•
9.3K Posts
0
September 10th, 2008 16:00
Remove access to the disks for 1 server and you'll be fine.
When using block level storage (like a SAN (whether it's fibre channel, iscsi or sas)), servers can only share disk space if there's some software in place that regulates which server gets to access which resource (virtual disk) at which time.
If you don't have any such software in place, you'll corrupt the filesystem and it's expected to run into data corruption/loss and other issues (like needing to run checkdisk on every boot).
yw_sau
3 Posts
0
September 11th, 2008 11:00
oh, noted. Am really appreciate it and really thanks alot
Again, if the storage and both server configure into cluster mode and the issue will be resolve? i'm not sure how the other SQL team will configure the cluster anyway, maybe you can help me with furture info with the cluster setting like refference or link?
Dev Mgr
4 Operator
•
9.3K Posts
0
September 11th, 2008 12:00
If 2 servers share a disk without software in place to manage this, the following situation can/will happen; one server starts writing to the disk, as it doesn't know about the other server it assumes it's the sole owner of the disk and can write to it as it pleases. Now the other server 'decides' to write to the disk as well. It doesn't have to 're-poll' what the latest state of the disk is (which data is where), as it's the only server using the disk in it's mind. So what happens is that this 2nd server can/will end up overwriting the data that the first server just wrote to the disk. This means you just lost the data that the first server wrote.
To set up a Microsoft cluster, there are a few requirements:
- all nodes that'll be in the cluster need to be in the same domain
- all nodes need a static IP address (I think you can force DHCP, but clustering really doesn't like this, even if you use reservations on your DHCP server)
- all nodes need to be running Windows 2000 Advanced Server, Windows 2003 Enterprise Edition, or Windows 2008 Enterprise Edition (or a Data Center version of these versions). NOTE: the MD3000 isn't certified for Windows 2000 clustering, but these OSes are for 'general clustering requirements'.
- the servers need to be running the same architecture (all 32-bit, or all x64; no mixes)
- highly recommended is to have a NIC in each server that can be dedicated to a heartbeat (private link between the servers). If it'll just be a 2-node cluster, this can be a patch/crossover cable. This will need a unique subnet (different from your regular network) with static IP addresses.
- You'll need a Quorum disk on the shared storage solution. This can be a regular data-disk, but as the quorum disk has certain limitations (e.g. if I'm not mistaken volume shadow copy cannot be done on the Quorum disk), it's recommended to just make a 1GB virtual disk on the MD3000 and reserve this for the Quorum (most people assign it drive letter Q, but this isn't a requirement).
A few notes about clustering:
- only 1 server can access a disk at a time
- if you want all (cluster) servers to be 'doing' something, you'll need to make multiple virtual disks and in the cluster create multiple cluster groups; each group will have it's own disk (if needed), own IP (if needed), and own DNS name (if needed).
yw_sau
3 Posts
0
September 12th, 2008 13:00
okie, noted and thanks for the reply
really appreaciate it :)