Re: client can't connect when instance on one particular node

We got it figured out - it was the named pipe thing mentioned
above....but just to followup - on that particular cluster we were
using 1433 for one instance, and 1434 for the other. But the only
workstations that were affected were the ones with named pipes above
tcp/ip in their preferred connection protocol on the client.

On Feb 3, 7:29 pm, "Anthony Thomas" <ALTho...@xxxxxxxxx> wrote:
What ports are defined for each of the SQL Server installations? Did you
specify a hard port number or are you using dynamic ports?

Best practice is to specify a specific port and then code the client to
direct to that port.

If you have the client telnet to the virtual ip and port, it should open to
a blank screen, which indicates a successful connection. If done
incorrectly, telnet will report a failed connection attempt.

Sincerely,

Anthony Thomas

--

<unc27...@xxxxxxxxx> wrote in message

news:1170373222.776799.316260@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

OK - so here's the setup - two node cluster (we'll say Node1 & Node2),
two instances (so multi instance, we'll say SQL1 & SQL2). W2K3 server
on the nodes, SQL 2000 SP4 for SQL version. Our situation is this.
We removed Node1 from the virtual server using the setup utility in
SQL 2000 in order to do some maintenance. Then server guys evicted
Node1 from the cluster, and we had SQL1 & SQL2 running on Node2 during
maintenance period. Server guys rebuilt Node1 (to boot to the SAN
instead of local drives) and then readded the server back into the
cluster configuration. At that point I readded the node back to the
SQL virtual server, and added SP4 as well - all appeared well. I put
SQL1 onto Node1 and SQL2 onto Node2.

But then we get a call from a (only one) customer saying that she
cannot connect through their application (this particular app connects
to SQL1). After some troubleshooting, I realize that she can only
connect when SQL1 resources are running on Node2. When I failover to
Node1, she can no longer connect. She can ping the virtual IP & the
DNS name that the application uses (the dns points to the virtual IP
of SQL1). Others in her environment can connect fine through the same
app. She can go to other machines & login with her login (SQL auth.)
& she can connect. So clearly it's a workstation issue - but it also
clearly seems to be related to the cluster, since this only happens
when I fail back to Node1, and it only happened after we did this
maintenance.

We're puzzled & need help. Any thoughts?


.

Relevant Pages

  • Re: N+1 Configuration
    ... Jasper Smith (SQL Server MVP) ... But looking in the cluster admin tool, ... mounting all other drives under that drive ...
    (microsoft.public.sqlserver.clustering)
  • Re: Clustering Question
    ... I have worked with Windows 2003 Small Business Server, Exchange Server 2003 and 2000 SQL server but not in a clustered environment. ... Do you want to cluster SQL? ... > you cannot install Cluster on a Small Business Edition, ...
    (microsoft.public.windows.server.clustering)
  • Re: Clustering Question
    ... The SQL Cluster would need to be ... SQL will need to Enterprise as well. ... So if I understand you correctly one computer will need Windows Server ...
    (microsoft.public.windows.server.clustering)
  • Re: Issues creating a new failover cluster on the same server
    ... G:\Microsoft SQL Server\MSSQL.1 ... CHNSQL04 Cluster Group ... Drives: H and F ... Group, the MS DTC Group, and each of the SQL Server Instance Groups running ...
    (microsoft.public.sqlserver.clustering)
  • RE: pen-test on a windows 2003 server box whit MS-SQL and Terminal Services
    ... laptop/pen-test system in as a Man-in-the-Middle between the SQL server ... > MS-SQL and Terminal Services ... > Have you done UDP port scans? ... obtaining the SQL version; since the OS is Win3k the SQL ...
    (Pen-Test)