October 5, 2010 at 10:55 am
I have a server hopelessly stuck on 2000 and have run into the following:
3 separate arrays:
OS on 1st array
Logs on 2nd array
Data and tempdb on 3rd array
No problems performance wise until 1 week ago. Ran profiler and saw very high duration times. Ran perfmon and saw disk pressure on the Data drive. Have done a number of things including reindexing (dropping and re-creating), updating all statistics, physical defrag of drives. Today when I run a profiler I am getting back reads and writes of all 0's which obviously can't be. Should have stellar performance, but still the same disk pressure when running perfmon. Stumped on this one. System logs don't indicate anything unusual with the server. Any ideas?
October 5, 2010 at 11:28 am
Sounds like unexpected outside pressure. I assume your drives are on a SAN. Get your network administrators to double check if your LUNs were recently shared with another project if you didn't set them up as dedicated spindles to the SQL server.
Never stop learning, even if it hurts. Ego bruises are practically mandatory as you learn unless you've never risked enough to make a mistake.
For better assistance in answering your questions[/url] | Forum Netiquette
For index/tuning help, follow these directions.[/url] |Tally Tables[/url]
Twitter: @AnyWayDBA
October 5, 2010 at 1:33 pm
Not on a SAN. All the drives are internal SCSI 15K. What is unusual is this pretty much happened overnight. Dedicated SQL Server so nothing else is on that server.
October 5, 2010 at 2:19 pm
I'd still be leaning towards a drive problem. What kind of arrays are they, and do you have a drive failure if they're RAID 5?
Never stop learning, even if it hurts. Ego bruises are practically mandatory as you learn unless you've never risked enough to make a mistake.
For better assistance in answering your questions[/url] | Forum Netiquette
For index/tuning help, follow these directions.[/url] |Tally Tables[/url]
Twitter: @AnyWayDBA
October 5, 2010 at 2:23 pm
They are what HP refers to as RAID 1+0, so 3 separate arrays. None of the drives are showing up as bad.
October 5, 2010 at 2:36 pm
Hrm. So, dedicated box, dedicated drives, no failures and all RAID 10. You've reindexed, defragged, and anything else that be causing physical churn.
You're getting spotty results between the perfmon and the db info, where it appears one isn't detecting properly. This occurred overnight, and no maintenance has recovered the issue.
Hrm. Double Hrm.
A couple of questions.
1) Have you done drive speed testing outside of SQL Server and gotten benchmarks prior to now? IE: Drop a largish data file to local, see how long it took. If so, have you repeated this after the problem to see if it's recurring. (Trying to determine if it's SQL, or physical).
2) Do most of your DBs exist in RAM? Do you currently have SQL server set to use all RAM on the box, and not leave anything for the OS. Are we dealing with pagefile churn or are these statistics only from the data array?
3) Did you recently have a brownout/etc? I'm wondering if a RAM chip went. Don't ask me how we troubleshot to it, but the end result was something similar. Where the server used to deal with memory for most of the issues, suddenly it was going to disk, a LOT. The RAM had popped on us.
I'm... out of ideas after that, unless there's some hidden detail I've missed or there's a surprising answer to an above question.
Never stop learning, even if it hurts. Ego bruises are practically mandatory as you learn unless you've never risked enough to make a mistake.
For better assistance in answering your questions[/url] | Forum Netiquette
For index/tuning help, follow these directions.[/url] |Tally Tables[/url]
Twitter: @AnyWayDBA
October 5, 2010 at 2:39 pm
Thanks so much. One thing I didn't think about was memory and it should have been my first thought. This server has rebooted on it's own with possible memory issues as the culprit, but never proved.
October 5, 2010 at 3:38 pm
SuzDBA (10/5/2010)
Thanks so much. One thing I didn't think about was memory and it should have been my first thought. This server has rebooted on it's own with possible memory issues as the culprit, but never proved.
No problem and my pleasure. Please post back with whatever solution you may find, or if you get more details. You've got me curious, especially about the memory settings and OS paging.
Never stop learning, even if it hurts. Ego bruises are practically mandatory as you learn unless you've never risked enough to make a mistake.
For better assistance in answering your questions[/url] | Forum Netiquette
For index/tuning help, follow these directions.[/url] |Tally Tables[/url]
Twitter: @AnyWayDBA
Viewing 8 posts - 1 through 7 (of 7 total)
You must be logged in to reply to this topic. Login to reply