Vertica standalone with RHEL cluster

hi,
i have 2 nodes running Veritca DB. i am using corosync + pacemaker for redundancy between 2 nodes. for DB sync i am using DRBD. which vertica services i need to manage in order to get it working?
thnaks

Comments

  • Hi Yaron,

    Vertica is a fully distributed database on its own, and is a single standalone service.  It does not need, and will not generally make use of, any of these applications.

    In order to get Vertica working, please just run the Vertica installer and specify the machines that should be configured as part of your cluster.  Let Vertica manage its own configuraion, and give it storage locations on local disk (not on DRBD partitions).

    (If you're deliberately not doing this, could you describe what you're doing in more detail?)

    Adam
  • Hi Adam Let me try to explain what i am trying to do. I have 2 nodes with vertica db installed as standalone mode , the installation is according to your procedures , with that I don't have any problem . When I am creating my db, it's working on each node without any problems . What I want to do is to sync both DB all time between both nodes. Thanks Yaron
  • Hi Yaron,

    Could you clarify what you mean by "standalone mode"?

    If you have correctly set up the two machines as a single cluster (by passing both of their public IP addresses in at the same time as arguments to a single call to "install_vertica"), then Vertica will automatically keep all data in sync.

    If you are not seeing this, then you have not followed our procedures correctly.

    Adam
  • Ok so just to make sure I got it right During installation I will provide both ips for both nodes. So the db will be sync between them and if I one node fails I will still be able to work?
  • Hi Yaron, You are correct that, if you provide both IPs during installation, you will get data synced between both nodes. Unfortunately, 2 nodes is too small for Vertica's high-availability functionality. You need at least three to survive nodes going down. Why? Well, what if a switch between the nodes goes down? Then each node thinks that the other is down, when in reality they are both up. If they each keep running, the databases' contents may diverge, and it may become impossible to merge them again without corrupting one of the two databases. Vertica will shut down (cleanly) rather than allow this sort of data corruption. (With three nodes, we can tell if one node goes down because the other two can still see each other.) Two nodes is small for a production Vertica DB. (Our free community license covers up to three.) We do have customers running production 2-node clusters. But they come with a bunch of headaches; we're really designed for workloads that want to scale out more than that. If you want access to Vertica's full functionality and performance, I'd strongly recommend three cheaper nodes instead. Adam
  • Hi Adam I understand the limitation but in our small scale systems we get only two nodes. Can I configure some 3rd device for example a switch ? By the way, for our large scale we are using 3 nodes

Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Emoji
Image
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file