MC/Service Gaurd - Clustering

[David_492]

MC/Service Guard - Clustering:


Dear all

I have one HP UNIX Server rp8400 which has 8 CPU and 4 Hard Disk 36GB and we partioned the server into two nodes node1 and node2 hardware partion (npar).


And I want to make cluster between these nodes node1 & node2 but I do not have any external disks or storage .how I can make cluster between these nodes even if I do not have cluster lock?

Is there any way I can work around to do the cluster

Thank You All

[Thayanidhi]

Hi Davind,

Clustering requires comman storage. Without which cluster doesn't have any
meaning. Also It's not worth doing cluster within the same box.
Cluster should be planned properly for SPF(single point of failure).

TT

Note: You have not assigned points for your more than 87 reposnses.
Please take some time to review the replies and award accordingly.
Also visit http://66.34.90.71/ITRCForumsEtiquette/index.html

[Sarvesh A Dhuri]

Hi !

Clustering using MC serive guard allows you to share common volume groups. I fyou dont have that then there is no point in going for cluster using Mc service guard.

Please go for VA7100 ,its is relatively cost effective

[Michael Steele_2]

In two-node clusters, a cluster lock is required. In three nodes it is suggested.

A cluster lock disk is used to prevent the possibility of a split brain cluster. Split brain is where the two nodes both come up owning the cluster, both become the primary and neither is an alternate. So a tie breaker is required. The tie-breaker is the cluster lock disk.

http://docs.hp.com/cgi-bin/fsearch/framedisplay?top=/hpux/onlinedocs/B5158-90018/B5158-90018_top.html&con=/hpux/onlinedocs/B5158-90018/00/00/16-con.html&toc=/hpux/onlinedocs/B5158-90018/00/00/16-toc.html&searchterms=cluster%7cquorum&queryid=20031108-073515

Even a quorum server, which performs the same functionality as a cluster lock disk needs to be independent:

http://docs.hp.com/cgi-bin/fsearch/framedisplay?top=/hpux/onlinedocs/B3936-90070/B3936-90070_top.html&con=/hpux/onlinedocs/B3936-90070/00/00/8-con.html&toc=/hpux/onlinedocs/B3936-90070/00/00/8-toc.html&searchterms=cluster%7cquorum&queryid=20031108-073515

You also need to investigate Single Points of Failure within the rp8400, or MC/ServiceGuard's High Availability design will be compromised:

http://forums1.itrc.hp.com/service/forums/parseCurl.do?CURL=%2Fcm%2FQuestionAnswer%2F1%2C%2C0x2daa3a7b3682d611abdb0090277a778c%2C00.html&admit=716493758+1068302606623+28353475