SQL Server Performance Forum – Threads Archive
SQL Server 2000 Cluster, Defragging and chkdsk /F
I was wondering if you might have a moment to answer the following question<br />which has plagued my very soul now for some time <img src=’/community/emoticons/emotion-1.gif’ alt=’
‘ /><br /><br />We are running a sql server 2000 cluster, with a shared E: drive which fails over should one of the servers be failed over – great! I was wondering if you knew of anyway to run chkdsk /F on this disk?<br /><br />My problem is obviously that when we fail one server over to perform some maintenance (so we’re not maintaining the active server effecting performance etc) the E: drive goes with it, so we then cant do simple things like a defrag or chkdsk…<br /><br />This leaves us with having to defrag the 100mb disc on the active server which obviously we’d rather not do.<br /><br />The disc is raid 5 I believe, not sure whether this makes much difference, a colleague suggested that we might not need to even bother with this kind of maintenance on a raided disc array…<br /><br />Any info you might have would be appreciated, and thank you in advance for your time reading my post.<br /><br />Regards<br /><br />Rob
We have never tried defragmenting disk in clustered environment, as never had such issue with defragmentation of files. I would wait for fellow peers opinion in this regard to use defrag software on a clustered environment or is it possible. Satya SKJ
Moderator
http://www.SQL-Server-Performance.Com/forum
This posting is provided “AS IS†with no rights for the sake of knowledge sharing.
We don’t defrag either. If you know the expected size of your databases and create the database files with this size (or more) from the beginning, there won’t be much physical fragmentation of the database files themselfes. If you really do want to run disk defrag and such you need to shut down the sql server service.
Hi, Thanks for the replies.. We’ve run defrag before without any problems – our 100gb disc has hardly any space left and the analysers shows it to be almost solid red (fragmented)…we have bought a new server and moved databases to this in a hope to save some disc space and increase performance on the main server. It was really the chkdsk /F thing that was the problem because obviously we couldn’t run that (from what we could see) without taking the entire server offline for a period of time, and as we work in the Health Service, we are not always able to just drop our applications/services like that. Part of our fragmentation could be because of our backups/log files which surely would be an issue for you all also? If you have these writing to the same disc (I know, I know…) and then they are automatically removed after say 3 days, this will most certainly lead to fragmentation.
Log files can also be set to a fixed size. For backup if you put them on a seperate drive that you don’t have SQL Data or Log files on you can run chkdsk /F etc on it.
We defrag all our databases on our clustered servers using Diskkeeper 8.0, and have never had any problems. Before doing so, many of our databases were very fragmented. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com
I use DiskKeeper 8.0 at home – its the mutts nutts, I have pursuaded the purchase of this at work also which is good – but again this is only covering the defrags and not the chkdsk’s etc.. Anyone know if a chkdsk *is* still required on a raid array? How does it work? Does it just see the disks as one disk like windows reports? Same with the defrag really…If I fire off a defrag on a raided disc array, and there’s say 10 physical discs in that array, does it do all of them?
Generally speaking, I prefer to use the utilities provided by the array vendor for "fixing" any potential issues. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com
So, would that be a "maybe run a defrag from time to time but other than that – leave it until it breaks" kinda thing?<br /><br />Just asking etc, as we are formalising our maintenance for each of the servers, and if there are ‘standards’ that typically people work to we’d like to do the same <img src=’/community/emoticons/emotion-1.gif’ alt=’
‘ /><br /><br />Typically the hardware would be maintained by our OPS people, who I guess in turn would contact our suppliers (Dell and so on)…
]]>