Cluster (active/active) and quorum
-
Hi Everyone,
I am working in a active/active cluster enviroment.
On node1 - i am receiving a ton of error messages indicating the clusvc cannot write to a file
"cluster service could not write to a file (c:\docume~1.....CLS56ED.tmp) the disk may be low on space or some other serious condition exisits
Event ID - 1080
My first thoughts are i have a permissions issue on the above mentioned location. But i have sense went in and given everyone full control of that folder. But the error still is being placed into the log.
Then I was looking at the node1 desktop, and went into the drives, and the Q drive is there, but not online (typical for cluster on the passive node) but in an active/active setup, should not both nodes see the quorum drive as being online?
I have done the cluster / quorum command on both nodes, and they both point to the Q:\MSCS location.
Im thinking the node1 is trying to write to the Q location, but cannot find it, so it puts the tmp files into the local storage section??
So my questions are this....
1) Does both nodes in an active/active setup need to have access to the quorum drive at all times?
thanks tony
-
How much space you have on your C drive?
-
I am assuming this is on Windows Server 2003 and not on 2008. If you are on 2008 then this may not be correct.
No, the quorom drive does not need to be accessible on all nodes in the cluster. The quorom drive is going to be a defined resource in the cluster group and hosted on whichever node is currently hosting the cluster group.
If the cluster group moves to the other node, then that node will own the drive and it will be accessible on that node.
If you are having problems with accessing the quorom drive from the node that is not hosting the cluster group, my guess is that the cluster has not been setup/configured correctly. But, that is just a guess and I could be wrong.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs -
From the small amount of information given, it looks like the cluster service is trying to write to its \Temp folder in the cluster service profile. Can you confirm if this is the case?
Can you see any of the files that are shown in the error log? Can you open any of them in Notepad and see what they contain?
I suspect something is wrong in the way your cluster is configured, as this is noit a problem I have seen on an Active/Active cluster myself.
Original author: https://github.com/SQL-FineBuild/Common/wiki/ 1-click install and best practice configuration of SQL Server 2019, 2017 2016, 2014, 2012, 2008 R2, 2008 and 2005.
When I give food to the poor they call me a saint. When I ask why they are poor they call me a communist - Archbishop Hélder Câmara
-
Tony, it would be nice if you could reply to the messages!
-
Thank you everyone for the responses.
Since both nodes have a quorum assignment to the drive letter Q, but only one active node can control the Q drive. The other node is trying to write to the Q drive, but it is not available to it, so it is trying to write to the temp drive as indicated in my first post.
I have tried creating a active/active cluster in virtual machines, with a nice little SAN, sql server still tries to use the Q drive for both nodes, even though only one node can access it, the other node cannot see or use the quorum.
Since this the same problem on both the real machines, and the virtual machines, and i did not setup the first cluster, i must make the huge assumption that just another MS bug in a rushed to release version of SQL Server.
So - if anyone has a magic wand that can correct this, please let me know.
So to answer everyones questions....
Yes, there is 10 gig available on both nodes C drive, i have given everyone full control of the directory that is being referenced in the errors.
take care
tony
-
In the Microsoft world, clustering is mainly intended to be a high-availability solution, not a DR solution. You only get a degree of DR protection if you have geographically dispered clusters.
Active/Active clustering is a way of making maximum use out of the hardware dedicated to a cluster. You have SQL cluster 1 running instance 1 on node 1, and SQL cluster 2 running instance 2 on node 2. Each SQL instance has exclusive use of its own databases. Each SQL instance can fail over to the other node if the original node fails. A typical use of this approach is for SQL replication, where the publisher runs on node 1 and the distributor runs on node 2 - this is what we have done for many years.
Underneath the SQL clustering is the Windows cluster. The Windows cluster uses the Quorum facility to help decide which nodes are running. If you use a disk as the quorum device, it is intended that this is online to only 1 node.
If you have configured your system so that more than one node wants to access the quorum at one time, then you have made a mistake in your configuration. If you do not understand why this is a mistake, then please take time to read more about Windows clustering. If you try to do what Windows clustering can deliver, then you will get a working system.
Original author: https://github.com/SQL-FineBuild/Common/wiki/ 1-click install and best practice configuration of SQL Server 2019, 2017 2016, 2014, 2012, 2008 R2, 2008 and 2005.
When I give food to the poor they call me a saint. When I ask why they are poor they call me a communist - Archbishop Hélder Câmara
-
Thanks for the quick lesson. This is not a DR setup, this was/is a kludge that was accomplished prior to my arriving, so I cannot give details on how or why it was done this way - simply that the reason it was done was because one of the databases was running slowly, and the second node was doing nothing (we know its running as hard as the active node, the IT great brains see 1% utilization as an underused device, so make it active and let users access it)
anyway - that is some back story.... new cluster being built up as we type.
thank you
-
Tony,I have been working with SQL Server cluster on Windows 2003 clusters for a bit more than 5 years and I never heard that every node needs to see the quorum. If you start the cluadmin utility (type "cluadmin") you will see your cluster divided in groups, one of the groups will be the cluster group (to which the quorum disk belongs). Every cluster group can be own by one node at a time. That is why when your quorum disk is visible on one of the nodes, the other nodes will not see it. I'm curious about what will happend if you move the cluster group to the node where you are seing the errors. If you can give it a try.
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply