Failed SQL Cluster Install Advice

A quick rant - After days of study, hours of install, and some killer
directions for Brad M McGehee I began to install a SQL Cluster on a Windows
2003 enterprise server cluster. If only i had found that one little article
that said that the install fails if you use two Cds.
So here I am for advice. My SQL 2005 cluster install failed when I
installed the second "tools" CD. Now I wonder what is the best resolution
for recovery. What I have read says that installing over a failed install is
not a good idea but in this case do i really have a failed install. The
installtion log shows no failures and everything that had installed up to
that point on both nodes had green "good" checks.
It states to install the installation files from both disks to a folder on
C: I assume that means the C: drive on the active node I used for the
installation would work right?
How should I proceed?
1. Try a standard unistall from control panel of all components on both
nodes.
2. Try to reinstall over the failed nut perhaps not failed install?
3. Complete the manual unistall as many artilces direct?
4. Any other proven ideas?
5. Send it to Bill and have him fix it. ha ha
Thanks for the advice ahead of time.
Regards, emagidson
The issue with installing from CD is that the remote node cannot direct the
local node to change the disk. By copying both CDs to a common location,
you eliminate the need to change disks. The exact location is not
necessarily important.
Don't do a reinstall on top of a broken install. Go to Control Panel |
Add/Remov programs and restart the existing install from there.
Note that some components such as client tools and SSIS do not install on
the remot enode during a cluster install. You have to install them on each
node locally. Personally, I prefer to do that install first, just to keep
things simple.
If the install restart fails, then you can try to do a manual remove and
re-install.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
news:1695A1A1-CE3B-4B61-BA66-79BC6A9874E5@.microsoft.com...
>A quick rant - After days of study, hours of install, and some killer
> directions for Brad M McGehee I began to install a SQL Cluster on a
> Windows
> 2003 enterprise server cluster. If only i had found that one little
> article
> that said that the install fails if you use two Cds.
> So here I am for advice. My SQL 2005 cluster install failed when I
> installed the second "tools" CD. Now I wonder what is the best resolution
> for recovery. What I have read says that installing over a failed install
> is
> not a good idea but in this case do i really have a failed install. The
> installtion log shows no failures and everything that had installed up to
> that point on both nodes had green "good" checks.
> It states to install the installation files from both disks to a folder on
> C: I assume that means the C: drive on the active node I used for the
> installation would work right?
> How should I proceed?
> 1. Try a standard unistall from control panel of all components on both
> nodes.
> 2. Try to reinstall over the failed nut perhaps not failed install?
> 3. Complete the manual unistall as many artilces direct?
> 4. Any other proven ideas?
> 5. Send it to Bill and have him fix it. ha ha
> Thanks for the advice ahead of time.
> Regards, emagidson
|||Geoff:
Thanks for the advice I do have a couple of questions. According to what I
read and since the install takes place from the active node, can i create a
folder on the root of node 1 C: drive say c:\sqlinstall then a folder under
that for both disks or do I need to do this on one of the shared array drives?
Also you mention restarting the install from control panel / add/remove
programs. This is not something I have done before. Again, longer term net
admin but I am not a clustering guy. Can you elaborate a bit on how to do it
from there given that I am seeing four to five "programs" that installed
through the install. We are installing everything via the install Sql,
Analysis tools, reporting service etc.
Also you mention other items that do not get installed, according to what I
read everything does get installed to the second node later in the process
and it looked like that was the case.
thanks again for the advice and in advance for more advice.
Regards, emagidson
"Geoff N. Hiten" wrote:

> The issue with installing from CD is that the remote node cannot direct the
> local node to change the disk. By copying both CDs to a common location,
> you eliminate the need to change disks. The exact location is not
> necessarily important.
> Don't do a reinstall on top of a broken install. Go to Control Panel |
> Add/Remov programs and restart the existing install from there.
> Note that some components such as client tools and SSIS do not install on
> the remot enode during a cluster install. You have to install them on each
> node locally. Personally, I prefer to do that install first, just to keep
> things simple.
> If the install restart fails, then you can try to do a manual remove and
> re-install.
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
>
> "EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
> news:1695A1A1-CE3B-4B61-BA66-79BC6A9874E5@.microsoft.com...
>
|||Local disk is best. Shared is NOT a good idea. You don't want the source
files going offline during the install or moving to another node.
The program you want to update is most likely called "Microsoft SQL Server".
That entry should have a "change" option. (I am working from memory here
since my Vista machine has very different labels for these functions.). The
installer will remember the earlier failure and try to correct it once you
have fixed the underlying issue.
Management Studio only gets installed to the local node during a cluster
install. Only items that have the "create failover cluster" option get
installed on the other node(s) in a cluster.
"EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
news:68B0121D-0FE1-434C-87FF-1C8B5A811676@.microsoft.com...
> Geoff:
> Thanks for the advice I do have a couple of questions. According to what
> I
> read and since the install takes place from the active node, can i create
> a
> folder on the root of node 1 C: drive say c:\sqlinstall then a folder
> under
> that for both disks or do
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
I need to do this on one of the shared array drives?[vbcol=seagreen]
> Also you mention restarting the install from control panel / add/remove
> programs. This is not something I have done before. Again, longer term
> net
> admin but I am not a clustering guy. Can you elaborate a bit on how to do
> it
> from there given that I am seeing four to five "programs" that installed
> through the install. We are installing everything via the install Sql,
> Analysis tools, reporting service etc.
> Also you mention other items that do not get installed, according to what
> I
> read everything does get installed to the second node later in the process
> and it looked like that was the case.
> thanks again for the advice and in advance for more advice.
> Regards, emagidson
> "Geoff N. Hiten" wrote:
|||Here is more information:
We completed a manual install as directed, changed the registry keys, even
killed the SQL registry entires. We have done this twice and both time get
the same message upon reinstall. "There is already a SQL instance with that
name". At this point I gues it is best to rebuild the whole thing as start
over. Unless someone else has a better idea.
We removed all of SQL from boht servers, everything on the shared array, the
folder that was still in Program files after the ininstall (come on uninstall
is uninstall get rid of everything) and the registry entires.
ALL OF THIS BECAUSE WE INSTALLED WITH TWO CDs WHICH DOES NOT WORK!!!!
Yes sorry I am very frustrated.
Thanks in advance, emagidson
"EMagidson" wrote:

> A quick rant - After days of study, hours of install, and some killer
> directions for Brad M McGehee I began to install a SQL Cluster on a Windows
> 2003 enterprise server cluster. If only i had found that one little article
> that said that the install fails if you use two Cds.
> So here I am for advice. My SQL 2005 cluster install failed when I
> installed the second "tools" CD. Now I wonder what is the best resolution
> for recovery. What I have read says that installing over a failed install is
> not a good idea but in this case do i really have a failed install. The
> installtion log shows no failures and everything that had installed up to
> that point on both nodes had green "good" checks.
> It states to install the installation files from both disks to a folder on
> C: I assume that means the C: drive on the active node I used for the
> installation would work right?
> How should I proceed?
> 1. Try a standard unistall from control panel of all components on both
> nodes.
> 2. Try to reinstall over the failed nut perhaps not failed install?
> 3. Complete the manual unistall as many artilces direct?
> 4. Any other proven ideas?
> 5. Send it to Bill and have him fix it. ha ha
> Thanks for the advice ahead of time.
> Regards, emagidson
|||Did you follow this KB article:
http://support.microsoft.com/kb/909967
(Perhaps the most complex KB article ever, but it should work. Just make
sure and perform the steps on both nodes.)
Also, remove any DNS and AD entries refering to the clustered instance.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
news:3763668C-691E-4DE0-871E-B450E36B7A68@.microsoft.com...[vbcol=seagreen]
> Here is more information:
> We completed a manual install as directed, changed the registry keys, even
> killed the SQL registry entires. We have done this twice and both time
> get
> the same message upon reinstall. "There is already a SQL instance with
> that
> name". At this point I gues it is best to rebuild the whole thing as start
> over. Unless someone else has a better idea.
> We removed all of SQL from boht servers, everything on the shared array,
> the
> folder that was still in Program files after the ininstall (come on
> uninstall
> is uninstall get rid of everything) and the registry entires.
> ALL OF THIS BECAUSE WE INSTALLED WITH TWO CDs WHICH DOES NOT WORK!!!!
> Yes sorry I am very frustrated.
> Thanks in advance, emagidson
> "EMagidson" wrote:
|||Geoff:
Thanks Geoff, We have deicded that we just don't feel we can trust the
systems with all the changes and have decided to install everything from
scratch. This way we know when out client begins to rely on the systems the
will be rock solid and ready to perform. Thanks for the link however I have
filed it into me personal knowledgebase.
Thanks again, emagidson
"Geoff N. Hiten" wrote:

> Did you follow this KB article:
> http://support.microsoft.com/kb/909967
> (Perhaps the most complex KB article ever, but it should work. Just make
> sure and perform the steps on both nodes.)
> Also, remove any DNS and AD entries refering to the clustered instance.
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
>
> "EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
> news:3763668C-691E-4DE0-871E-B450E36B7A68@.microsoft.com...
>
|||Good call. I would recommend the same thing in the same situation. Clean
systems are generally the best, especially in a complex, clustered
environment.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
news:C762EE7E-68A2-428C-8B2F-55F99BBE6555@.microsoft.com...[vbcol=seagreen]
> Geoff:
> Thanks Geoff, We have deicded that we just don't feel we can trust the
> systems with all the changes and have decided to install everything from
> scratch. This way we know when out client begins to rely on the systems
> the
> will be rock solid and ready to perform. Thanks for the link however I
> have
> filed it into me personal knowledgebase.
> Thanks again, emagidson
> "Geoff N. Hiten" wrote:
|||Geoff:
As you can most likely tell I am not a newby to IT and networks but I am
with Clustering. I wanted to thank you for all your help. We have now
reinstalled the whole shibang and have everything running. All of the
failover tests have been performed and the only thing left to do is create
backups of the servers.
I will be beginning a new thread to ask a few more basic questions "Basic
SQL cluster questions" that if you have time I would love to hear your
replies.
Again thank you. I always make a point to help out in newsgroups as much as
I can. It always seems to come back in spades.
Regards, emagidson
"Geoff N. Hiten" wrote:

> Good call. I would recommend the same thing in the same situation. Clean
> systems are generally the best, especially in a complex, clustered
> environment.
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
>
> "EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
> news:C762EE7E-68A2-428C-8B2F-55F99BBE6555@.microsoft.com...
>
|||You are welcome.
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
"EMagidson" <EMagidson@.discussions.microsoft.com> wrote in message
news:96BC8AFB-F754-43FD-A412-137B8C3EE1F9@.microsoft.com...[vbcol=seagreen]
> Geoff:
> As you can most likely tell I am not a newby to IT and networks but I am
> with Clustering. I wanted to thank you for all your help. We have now
> reinstalled the whole shibang and have everything running. All of the
> failover tests have been performed and the only thing left to do is create
> backups of the servers.
> I will be beginning a new thread to ask a few more basic questions "Basic
> SQL cluster questions" that if you have time I would love to hear your
> replies.
> Again thank you. I always make a point to help out in newsgroups as much
> as
> I can. It always seems to come back in spades.
> Regards, emagidson
> "Geoff N. Hiten" wrote:
sql

Failed Maintenance Plan

Please advice.
I run a SQL2k SP4 on WIN2K SP4.
I have a maintenance plan to backup the databases at 01:30AM. This has alway
s been succesfull.
Strangely it did the backup at 02:25PM with no entry in SQL Error Log and ca
using subsequent backups to fail as the files created had the 02:25 badge he
nce could not be deleted.
Please help as to what caused the maintenance plan to backup at 02:25 and wh
at to do to avoid it.
Thanks for your kind response.I posted a reply to this yesterday. Here's that text:
My guess is that it wasn't Main Plan that did the backup, someone else did i
t (or explicitly started the main
plan job, or scheduled the main plan job at that hour). If it was executed a
s a job, you can view the job
history under EM, Management, Jobs. In any event, each backup is written to
the eventlog and the errorlog
file.
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
"Timi" <anonymous@.discussions.microsoft.com> wrote in message
news:EA0A7329-EAA1-49B3-9DDB-1736B6F35A4C@.microsoft.com...
> Please advice.
> I run a SQL2k SP4 on WIN2K SP4.
> I have a maintenance plan to backup the databases at 01:30AM. This has alw
ays been succesfull.
> Strangely it did the backup at 02:25PM with no entry in SQL Error Log and causing
subsequent backups to fail
as the files created had the 02:25 badge hence could not be deleted.
> Please help as to what caused the maintenance plan to backup at 02:25 and
what to do to avoid it.
> Thanks for your kind response.
>|||Thanks for your response.
The maint plan was ran at 02:25AM as opposed to 01:30AM the scheduled time.|||can you check with your server guys if there was a problem
with the server which resulted in it being restarted on or
arounf 2:25 ?, but the best way of finding out is to check
the error log.
J

>--Original Message--
>Thanks for your response.
>The maint plan was ran at 02:25AM as opposed to 01:30AM
the scheduled time.
>.
>|||Agent does no run a job if it was down at the time it should have executed.
I doubt that this is the reason,
based on that. I bet a beer that someone started the job at that time.
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
"Julie" <anonymous@.discussions.microsoft.com> wrote in message news:13fdd01c41357$84085e80$
a301280a@.phx.gbl...
> can you check with your server guys if there was a problem
> with the server which resulted in it being restarted on or
> arounf 2:25 ?, but the best way of finding out is to check
> the error log.
> J
>
> the scheduled time.|||Bet a beer against you and your SQL Server knowledge, never ;)
Anyway your right but my sneaky plan was to get Timm to check the server, as
we both agree thats where the problem is, and who actually started it.|||:-)
<<as we both agree thats where the problem is, and who actually started it.>
>
Yep.
--
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
"Julie" <anonymous@.discussions.microsoft.com> wrote in message
news:D6CB3367-9F9F-4633-B7FE-3FBEBFFA8B74@.microsoft.com...
> Bet a beer against you and your SQL Server knowledge, never ;)
> Anyway your right but my sneaky plan was to get Timm to check the server, as we bo
th agree thats where the
problem is, and who actually started it.

Failed Maintenance Plan

Please advice.
I run a SQL2k SP4 on WIN2K SP4.
I have a maintenance plan to backup the databases at 01:30AM. This has alway
s been successfull.
Strangely it did the backup at 02:25PM with no entry in SQL Error Log and ca
using subsequent backups to fail as the files created had the 02:25 badge he
nce could not be deleted.
Please help as to what caused the maintenece plan to backup at 02:25 and wha
t to do to avoid it.
Thanks for your kind response.My guess is that it wasn't Main Plan that did the backup, someone else did i
t (or explicitly started the main
plan job, or scheduled the main plan job at that hour). If it was executed a
s a job, you can view the job
history under EM, Management, Jobs. In any event, each backup is written to
the eventlog and the errorlog
file.
Tibor Karaszi, SQL Server MVP
http://www.karaszi.com/sqlserver/default.asp
"Timi" <anonymous@.discussions.microsoft.com> wrote in message
news:CDA33615-5054-47D3-8D06-0A52B8FB7544@.microsoft.com...
> Please advice.
> I run a SQL2k SP4 on WIN2K SP4.
> I have a maintenance plan to backup the databases at 01:30AM. This has alw
ays been successfull.
> Strangely it did the backup at 02:25PM with no entry in SQL Error Log and causing
subsequent backups to fail
as the files created had the 02:25 badge hence could not be deleted.
> Please help as to what caused the maintenece plan to backup at 02:25 and w
hat to do to avoid it.
> Thanks for your kind response.
>

Failed DTS Package locks user out of SQL2005 (Windows Authentication) ... for good!

Would be interested in any advice or comment on the issue we are experiencing with SQL 2005.

In order to test some DTS package migrations, we simply created a DTS package on SQL 2000 (using Export) to copy all database objects from one database to another.

Then we restored the source database on SQL 2005 server, migrated the DTS package, and tried to execute it.

The package migrated 'fine', ran part way, and then failed. Since then we have been unable to connect to the server (Database Engine, or Integration Services) using Windows Authentication. (Error 18456, Sev 16, State 11).

We can connect to the Database Engine using SQL Authentication (but not to Integration Services which only allows Windows Authentication).

We have been able to replicate this consistently - every annoying time we've tried it!

(Reinstalling SQL Server does 'fix' the issue ... until you try a similar package again).

Are you saying that you can't connect to Integration Services? From your post the Database Engine is fine, and 'locks user out' is misleading as the account is not locked out.

What does the package try or fail to do (and which step)?
What appears in the application event log when you fail to connect to IS?
Is IS service still running?|||

Thanks for your reply. Allow me to attempt to answer your queries.

Both Integration Services and Database Engine (MSSQLSERVER) are still running. As stated, we can connect to the Database Engine using SQL Authentication, but Windows Authentication (no matter what account) no longer works to either the Database Engine or Integration Services.

OK, perhaps my terminology is not quite correct - it's not an account lock-out, but that's pretty much the symptom. Perhaps I should say "Windows Authentication is broken"?

Sample Event Log (descending timestamp order ie. same as in event log viewer):

MSSQLSERVER Failure Audit (4) 18456 SERVER\Administrator Login failed for user 'SERVER\Administrator'. [CLIENT: <local machine>]

SQLISPackage Error None 12291 SERVER\Administrator SERVER "Package ""DTS_TEST"" failed."

SQLISPackage Error None 12291 SERVER\Administrator SERVER "Package ""ShellPackage"" failed."

SQLISPackage Information None 12288 SERVER\Administrator SERVER "Package ""ShellPackage"" started."

SQLISPackage Information None 12288 SERVER\Administrator SERVER "Package ""DTS_TEST"" started."

At this stage I can't determine which step in the package is broken, but that's not my biggest issue at this point. Why Windows Authentication "breaks", and how to fix it is my real concern.