Monday Feb 19, 2007

Confusing message of sc_delegated_restarter

Sun Cluster 3.2 provides a new daemon called sc_delegated_restarter as part of the normal running cluster framework on each node of the cluster. The sc_delegated_restarter is SMF (Service Management Facility of Solaris 10) and RGM (Resource Group Manager of Sun Cluster) aware. Therefore you can manage your existing SMF services with the provided resource types SUNW.Proxy_SMF_failover, SUNW.Proxy_SMF_multimaster or SUNW.Proxy_SMF_scalable of Sun Cluster 3.2.

Maybe you see the following message within the boot of your cluster which you can simple ignore.

Jan 29 08:40:16 host345 Cluster.SMF.DR: Unable to open door descriptor /var/run/rgmd_receptionist_door
Jan 29 08:40:16 host345 Cluster.SMF.DR:
Jan 29 08:40:16 host345 Error in scha_cluster_get

This is a race condition where sc_delegated restarter comes up before RGM.
This is the known Bug 6490541: The sc_delegated_restarter SMF svc should depend on rgm SMF svc

Update June 18 2007
Which is fixed in the Sun Cluster 3.2 core patch (125511-02 S10, 125512-02 S10_x86, 125510-02 S9)

Wednesday Feb 07, 2007

Start with Sun Cluster 3.2

The next powerful version of Sun Cluster is available. Sun Cluster 3.2 is part of the Solaris Cluster product suite which includes Sun Cluster, Sun Cluster Geographic Edition, developer tools and support for commercial and open-source applications through agents. Some resources I like to mention:

For the installation of Sun Cluster 3.2 you should know:
--> Solaris 10 11/06 (Update3) is required. No earlier release of Solaris 10 is supported.
--> Addtional the following patches are highly recommended. (The table lists minimum revision of the patches)

118833-36 SunOS 5.10: kernel patch
124918-02 SunOS 5.10: devfsadm, devlinks, drvconfig patch
124916-01 SunOS 5.10: sd, ssd drivers patch
121010-04 SunOS 5.10: rpc.metad patch
120986-10 SunOS 5.10: mkfs and newfs patch
118855-36 SunOS 5.10_x86: kernel patch
124920-02 SunOS 5.10_x86: Solaris boot filelist.ramdisk patch
124919-02 SunOS 5.10_x86: devfsadm patch support
124917-02 SunOS 5.10_x86: sd driver patch
121011-04 SunOS 5.10_x86: rpc.metad patch
120987-10 SunOS 5.10_x86: mkfs, newfs, other ufs utils patch

Note: Sun Cluster 3.2 can also run on Solaris 9 8/05 (Update8). Certainly all Solaris 10 specific features are not available in Solaris 9.

Sunday May 07, 2006

Patching to Sun Cluster 3.1 Update4

Patching a Sun Cluster 3.1 on Solaris 8 or Solaris 9 is not always straight forward. Especially if you install the Sun Cluster core patch which is equivalent to Sun Cluster 3.1 8/05 aka Update4 release.

The following patches are equivalent to Sun Cluster 3.1 8/05 aka Update4:
Solaris 9: 117949-12 or higher
Solaris 9_x86: 117909-12 or higher
Solaris 8: 117950-12 or higher

If you install a mentioned SC core patch without to shutdown the whole cluster then patching a node is similar to rolling upgrade. It is also called "Rebooting Patch (Node) procedure". For these procedures it is a requirement to have the -11 revision of the SC core patch installed and active. Active means you need to install -11 revision of the patch and reboot the nodes. If the -11 SC core patch is not installed AND active then you will hit the bug 6210440 and your whole cluster is going down within patch installation. The issue is documented on several places but still many Sun Cluster 3.1 will get affected. Therefore never forget to read the special install instructions of these patches.

Beware if you use TLP (Traffic Light Patchmanagement) only one revision of the SC core patch is in the patchset. If the patchset include the revision -12 or higher then you should have an active -11 SC core patch befor you install the TLP patchset.

It is not necessary to look to the Solaris 10 SC core patches because Sun Cluster 3.1 8/05 is the first release which support Solaris 10. But I like to mention, it is highly recommended to use Solaris 10 1/06 aka Update1 for Sun Cluster 3.1 8/05 on Solaris 10.

Thursday Mar 30, 2006

Quota with Sun Cluster

Do you like to use quota with Sun Cluster 3.1?

First of all take care of the following restrictions:

  • You can use quota only within SUNW.HAStoragePlus resource.

  • You can use it only with a failover filesystem (FFS).

  • Quotas are not working on global mounted filesystems.

To configure the quota you need to add the mount option “quota” or “rq” to the FFS in the /etc/vfstab. Furthermore you need to configure the SUNW.HAStoragePlus for the FFS.

Simple example:

  1. Register the resource type
    # scrgadm -a -t SUNW.HAStoragePlus

  2. Create a resource group
    # scrgadm -a -g FFSq-rg

  3. Add filesystem to vfstab
    # vi /etc/vfstab (on all nodes)
    /dev/md/nfsset/dsk/d0 /dev/md/nfsset/rdsk/d0 /quotafs ufs 2 no logging,rq

  4. Create resource for HAStoragePlus
    # scrgadm -a -j hastp-rs -g FFSq-rg -t SUNW.HAStoragePlus -x FileSystemMountpoint=/quotafs -x AffinityOn=TRUE

  5. Switch the resource group online
    # scrgadm -Z -g FFSq-rg

  6. Configure the quotas
    # touch quotas
    # chmod 600 quotas
    # edquota -v USER
    # quotaon -av

  7. Test the switchover of the resource group
    # scswitch -z -g FFSq-rg -h

After the switch of the filesystem /quotafs the quota will be started automatically from the SUNW.HAStoragePlus resource. To configure your setup of the quota you can refer to the quota manpage.


I'm still mostly blogging around Solaris Cluster and support. Independently if for Sun Microsystems or Oracle. :-)


« April 2014