I have a large number of databases on a couple of heavily
utilized production 2000 servers (about 80 on one and
about 120 on another). Both servers hung mysteriously
last week and one of them had the following error:
"WARNING: EC 859eb5c8, 0 waited 300 sec. on latch a04e98.
Not a BUF latch."
The KB (310834) says this is from an overstressed system,
however my perf mon stats do not show this (unless context
switching greater > 5,000 for sustained periods is a
problem). But that same KB article says that
autogrow/autoshrink can be responsible and due to space
limitations, I have the autoshrink on all db's. So here
are my questions:
1. How can I monitor autogrow and autoshrink?
2. How can I tell how much cpu and disk they're using?
3. What is an acceptable threshold for context switching
for a four cpu server?
Why on earth do you have autoshrink enabled?
http://www.aspfaq.com/
(Reverse address to reply.)
> 1. How can I monitor autogrow and autoshrink?
> 2. How can I tell how much cpu and disk they're using?
> 3. What is an acceptable threshold for context switching
> for a four cpu server?
|||Space. They wouldn't get me more space so I kept having
drives filling up. I finally talked them into getting me
more space on one of the servers and just did a
migration. I of course am about take all the autoshrinks
off on that server.
But on the other server where I am still struggling with
space issues, I am looking for "fuel" to get them to get
me more space. That's part of why I'd like to monitor the
server and show them the effect (besides fragmentation of
course) that this is having on the server.
How can I show them that?
>--Original Message--
>Why on earth do you have autoshrink enabled?
>--
>http://www.aspfaq.com/
>(Reverse address to reply.)
>
>
>.
>
|||Well, you can have a trace running with minimal events (e.g.
autogrow/autoshrink) and capture the duration / cpu usage of those specific
events. Have you ever used Profiler?
http://www.aspfaq.com/
(Reverse address to reply.)
"Art" <anonymous@.discussions.microsoft.com> wrote in message
news:008f01c4a63b$41f93970$a301280a@.phx.gbl...[vbcol=seagreen]
> Space. They wouldn't get me more space so I kept having
> drives filling up. I finally talked them into getting me
> more space on one of the servers and just did a
> migration. I of course am about take all the autoshrinks
> off on that server.
> But on the other server where I am still struggling with
> space issues, I am looking for "fuel" to get them to get
> me more space. That's part of why I'd like to monitor the
> server and show them the effect (besides fragmentation of
> course) that this is having on the server.
> How can I show them that?
|||Here's some ammunition:
http://www.karaszi.com/SQLServer/info_dont_shrink.asp
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
http://www.solidqualitylearning.com/
"Art" <anonymous@.discussions.microsoft.com> wrote in message
news:008f01c4a63b$41f93970$a301280a@.phx.gbl...[vbcol=seagreen]
> Space. They wouldn't get me more space so I kept having
> drives filling up. I finally talked them into getting me
> more space on one of the servers and just did a
> migration. I of course am about take all the autoshrinks
> off on that server.
> But on the other server where I am still struggling with
> space issues, I am looking for "fuel" to get them to get
> me more space. That's part of why I'd like to monitor the
> server and show them the effect (besides fragmentation of
> course) that this is having on the server.
> How can I show them that?
Sunday, February 12, 2012
Autoshrink and Autogrow
Labels:
andabout,
autogrow,
autoshrink,
couple,
database,
databases,
heavilyutilized,
microsoft,
mysql,
number,
oracle,
production,
server,
servers,
sql
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment