Showing posts with label advanced. Show all posts
Showing posts with label advanced. Show all posts

Friday, February 24, 2012

Clustering with sql and third party database softwares

I have a clustered enviroment made of 2 win2k advanced servers
clustered in a network. Servers are running sql 2000 that hosts a third
party database X. Now X was working perfectly until these servers were
clustered. Users accessed X via a citrix client. Any idea as to why
accessing this third party database (X) became slower after being
placed in a clustered enviroment? Please feel free to come back with
any solutions.
Thanks
Hi
Did you have the exact same hardware setup before you clustered? A badly
implemented SAN can blow your performance.
Is each instance allocated the same amount of RAM as before clustering?
Run performance monitor and see what type of bottlenecks are shown. You did
not give much information about what is slower so I can't really pin point
the problem.
Regards
Mike Epprecht, Microsoft SQL Server MVP
Zurich, Switzerland
IM: mike@.epprecht.net
MVP Program: http://www.microsoft.com/mvp
Blog: http://www.msmvps.com/epprecht/
<dafeuwotu@.yahoo.co.uk> wrote in message
news:1121623805.177041.48990@.f14g2000cwb.googlegro ups.com...
>I have a clustered enviroment made of 2 win2k advanced servers
> clustered in a network. Servers are running sql 2000 that hosts a third
> party database X. Now X was working perfectly until these servers were
> clustered. Users accessed X via a citrix client. Any idea as to why
> accessing this third party database (X) became slower after being
> placed in a clustered enviroment? Please feel free to come back with
> any solutions.
> Thanks
>

Sunday, February 19, 2012

Clustering an existing set of servers

I have two non-clustered win2k servers (advanced server) with SQL Enterprise
installed with local storage. Production use has already started on a limited
basis. Would it be possible to to in place clustering using the following
steps?
1.Installation and configuration of additional hardware on each sever to
support SAN connectivity.
2.Getting the necessary IP addresses for one (active/passive) or two
3Installation and configuration of Microsoft Cluster Service on each server.
4.Install and configure named instance
5. Add the database logins from the non-clustered instances to
the newly created name instance.
6. Restore all users databases onto the new named instance.
Has anyone attempted this and been successful. The discussion of high
availability solutions including clustering was put off until just before
go-live. Hardware and budget are limited.
Thank you in advance for any suggestions
From a very high level, yes, the general steps are right but the details are
quite a bit more involved.
Before going into that, I strongly suggest you check out the MS clustering
hardware certification list. While you can hook up lots of different
hardware for a homegrown failover cluster solution, and many of them will
work just fine, if something bad happens, you'll have challenges with
Microsoft support. There are good reasons clusters are certified as a system
and not as "mix-n-match" components. Cluster hardware really isn't much more
expensive than "regular" stuff these days. If you're serious about HA,
convince your management/finance folks that it's not worth saving 10-15%
cost now and risk significant pains and costs later.
aK.
"John Kingery" <John Kingery@.discussions.microsoft.com> wrote in message
news:BAB4AD2D-B866-4EE2-AAE2-77856D7D84E7@.microsoft.com...
> I have two non-clustered win2k servers (advanced server) with SQL
Enterprise
> installed with local storage. Production use has already started on a
limited
> basis. Would it be possible to to in place clustering using the following
> steps?
> 1. Installation and configuration of additional hardware on each sever to
> support SAN connectivity.
> 2. Getting the necessary IP addresses for one (active/passive) or two
> 3 Installation and configuration of Microsoft Cluster Service on each
server.
> 4. Install and configure named instance
> 5. Add the database logins from the non-clustered instances to
> the newly created name instance.
> 6. Restore all users databases onto the new named instance.
> Has anyone attempted this and been successful. The discussion of high
> availability solutions including clustering was put off until just before
> go-live. Hardware and budget are limited.
> Thank you in advance for any suggestions
>
|||I am pretty sure you will have to uninstall SQL and reinstall for it to be
cluster aware, though I have never done this.
Even if it does work, would you trust your configuration? Take one of the
machines, attach external storage, start over and cluster it, move the
databases from the existing SQL using the attach command, when moved, redo
the old SQL server. That is what I would do
Cheers,
Rod
MVP - Windows Server - Clustering
http://www.nw-america.com - Clustering
http://msmvps.com/clustering - Blog
"John Kingery" <John Kingery@.discussions.microsoft.com> wrote in message
news:BAB4AD2D-B866-4EE2-AAE2-77856D7D84E7@.microsoft.com...
>I have two non-clustered win2k servers (advanced server) with SQL
>Enterprise
> installed with local storage. Production use has already started on a
> limited
> basis. Would it be possible to to in place clustering using the following
> steps?
> 1. Installation and configuration of additional hardware on each sever to
> support SAN connectivity.
> 2. Getting the necessary IP addresses for one (active/passive) or two
> 3 Installation and configuration of Microsoft Cluster Service on each
> server.
> 4. Install and configure named instance
> 5. Add the database logins from the non-clustered instances to
> the newly created name instance.
> 6. Restore all users databases onto the new named instance.
> Has anyone attempted this and been successful. The discussion of high
> availability solutions including clustering was put off until just before
> go-live. Hardware and budget are limited.
> Thank you in advance for any suggestions
>

Tuesday, February 14, 2012

Clustered SQL Server freezes & restarts itself in Enterprise Manager

Hi,
In our large organisation we have SQL Server 2000 running on a 2 node cluster using Windows 2000 Advanced Server Clustering. Each node has 2GB RAM. The server has about 10 databases only one of which is significant size at about 1GB. Our website accesses this database via ASP. The website receives between 10000-40000 hits per hour.
The problem is that periodically (not necessarily during high load periods at all) this single database 'freezes' (not the whole server - the other databases seem fine) and refuses all login attempts from the website. This happened again on the weekend when the database was down for about 6 hours. When we go into Enterprise Manager and click on the instance, there is initially no activity for an unusual amount of time but eventually the instance is selected and the tree expands. When this happens everything is ok again and the website can access the database again.
We have only just enabled failure auditing today, so we don't have any info from there and are at a loss as to whats going on. Has anyone out there had similar experiences or have any ideas?

(We will look at upgrading to 2005 and upgrading the machines too but being a big company that will take a lot of time with red tape etc.)

Many thanks.

Hello,

Hopefully not an offensive question, but is there anything in either the win eventlogs or sql errorlog?

Is the problematic databases' filegroup on a seperate drive (resource) to the other db's?

Cheers

Rob

|||Hi Rob,Thanks for the reply. No there was nothing suspicious in the SQL errorlog and nothing strange in the Windows event log either. All of the databases use the same shared storage.Cheers, Nathan.|||

Hi Nathan,

I would have leant towards either blocking or an inordinate amount of locks, however the latter would affect all databases on the instance. Is there a chance the outage is being caused by the db (auto)growing (either log or data)? I think this would only be presented if it were a "large" db...

In order to narrow it down a bit, I'd be create and run a perfmon log, with all the usual suspects: memory,page reads/faults, cpu, tps, log growth, cache hit ratio, user connections etc. Run the log until you encounter the db freeze, and hopefully the perfmon log will at least give you some clue.

Are you sure it's just the one db that freezes, or have you seen if affect others?

Cheers,

Rob