Application of quorum

Environment

2 RAC cluster nodes with each container:

----------------------------------------

Oracle Linux 6 update 5 (x 86-64)

Oracle Grid Infrastructure 12R1 (12.1.0.2.0)

12R1 database Oracle (12.1.0.2.0)

I'm not understanding the role of the College in the Group of lack of quorum for voting with Oracle ASM files.

Failure of groups (FG) assure that there is a separation of the risk you are trying to mitigate. The separation is to scale with user data. Vote files, the file is separated on a disk for each FG in the Group Disk (DG). If voting files are stored on a disk group (DG) to the normal dismissal. A minimum of two FGs is required. It is recommended that 3 FGs serve. It's so partner status Table (PST) may have at least one other FG where PST is maintain by comparison. It is in the case of a failure of FG. Must the FGs be QUORUM which store files to vote? What is the role of the College? When is it necessary?

Hello

I'll start with what the Quorum means:

The quorum is the minimum number of members (majority) of a game, necessary to avoid a failure. (COMPUTER concept)

A quorum is a lot like Votedisk Quorum Quorum OCR, network Quorum, Quorum PST, etc.

We must separate this quorum, we are concerned.

The PST Quorum is different from the Quorum of Votedisk, well think everything works toogheter.

Quorum PST:

A PST file contains information on all the disks in a Diskgroup ASM - number of disk, Disk Status, disk number, the heart rate info and Failgroup Info partner.

A disk group must be able to access a quorum of the Tables of status partner (PST) to mount the diskgroup.

When diskgroup mount is requested the instance reads all disks in the disk group to find and verify all available PST. Once it checks that there is enough PSTs for quorum, he set up the disk group.

There is a good post here: ASM support Guy: partnership and the State Table

Votedisk quorum:

Is a minimum number of votes for cluster is operational. There is still a quorum of votedisk.

When you configure votedisk in normal redundancy you 3 Votedisk one in each Failgroup. For cluster is operational you need of at least a quorum with 2 online voting for the cluster are online.

Quorum Failgroup (clause):

Quorum Failgroup is an implementation of a Diskgroup option.

This option do not confuse with the Quorum of voting, because the Quorum of voting and Quorum Failgroup are different things.

For example: in a Normal Redudancy diskgroup can I lost my Failgroup Quorum, and the Cluster will stay in line with 2 Failgroup regular, this Quorum failgroup is an installation program.

Named as 'Failgroup Quorum' a failegroup to a specific purpose which is store only the Votedisk due to deployment of the infrastructure of the Oracle.

Is no obligation to use "Quorum Failgroup' in a Diskgroup containing votedisk.

Now, back to your question:

If your failure groups have only 1 ASM disk then should not the recommendation to use high redundancy (5 groups of failure) so in the case of a disk failure ASM a quorum of PST (3 PST files) would be possible?

About Quorum PTS: You need must be aware that if you have 5 PST, you will need at least a quorum with 3 PST to mount his Diskgroup.

If you have 5 Failgroup and each Failgroup has ASMDISK that one alone, you'll have a STP by ASMDISK that support you lost at least 2 PST to make a Quorum with 3 PST and keep mounted Diskgroup or fix it.

The bold italic Oracle documentation above seems to say that if you allocate 3 devices in disc 2 will be used by groups of normal redundancy failure. In addition, a group of lack of quorum will exist who will use all disk devices. What does that mean?

I don't know what documents are saying is so confused. I'll try to contact some employee of Oracle to check it out.

But will attempt to clarify some things:

Suppose you install a Diskgroup as follows:

DATA DiskGroup

Failgroup data01

* / dev/hdisk1 and/dev/hdisk2

Failgroup data02

* / dev/hdisk3 and/dev/hdisk4

Failgroup data03

* / dev/hdisk5 and/dev/hdisk6

Quorum Failgroup data_quroum

* / nfs/votedisk

When you add Votedisk on this WHAT DATA diskgroup, the CSSD will store as follow:

CSSD will randomly choose an asmdisk by votedisk Failgroup and magazine on the subject, but always chooses an ASMDISK of Failgroup of Quorum (if it exists).

Therefore, after you add votedisk below diskgroup that you have:
* Failgroup data01 (/ dev/hdisk2)

* Failgroup data03 (/ dev/hdisk5)

* Failgroup data_quorum (votedisk/nfs /)

Mount diskgroup data, you must failgroup data01, data03 and data_quorum available in the diskgroup mount, otherwise diskgroup does not rise.

On Documentation (https://docs.oracle.com/database/121/CWADD/votocr.htm#CWADD91889) is a bit of confusion:

Normal redundancy

The level of redundancy you choose for the Oracle ASM disk group determines how Oracle ASM mirrors the disk group files and determines the number of disks and the amount of disk space you need. If the files with the right to vote in a disk group, groups of disks that contain Oracle Clusterware (OCR and vote files) files have a more minimum number of groups of failure than other disk groups because with voting rights are stored in the default quorum groups.

What he says is: when using Failgroup of Quorum, you will have a more minimal number of failure than other disk groups...

But remember that qorum Failgroup is optional for those who use a single storage or an odd number of H/w storage.

For Oracle Clusterware files, a normal redundancy disk group requires a minimum of three disk devices (two of the three discs are used by groups of failure and all three disks are used by the default quorum group) and provides three files with right to vote and a mirror of the OCR OCR. When you use a normal disk of redundancy group, the cluster can survive the loss of the group a failure.

Try to clarify:

-Votedisk in Diskgroup with normal redundancy requires three drive devices. (When using Failgroup Quorum: you will have two of the three disk used by regular failgroup and one of the three discs are used by Quorum Failgroup but all three discs (regular failgroup and quorum that store votedisk) count when mount the diskgroup.)

- OCR and a mirror of OCR:

It was really confusing. Because the mirror of the OCR must be placed in a different diskgroup because OCR is stored similar to how Oracle database files are stored. Scopes are spread across all disks in the diskgroup.

I don't know what he's talking about. If measure on diskgroup redundancy mirror or mirror of the OCR.

By as note and especially the documentation says it's not possible store OCR, and OCR mirror on diskgroup even

RAC FAQ (Doc ID 220970.1)

How the registry of Cluster Oracle (OCR) is retained when using ASM?

And (https://docs.oracle.com/database/121/CWADD/votocr.htm#CWADD90964)

* At least two places of OCR if OCR is configured on an Oracle ASM disk group. You must configure OCR in two groups of independent labels. In general, it is the recovery area and work space.

High redundancy:

For the Oracle Clusterware, a disk group files high redundancy requires a minimum of five disk drives (three of the five discs are used by groups of failure and all five disks are used by the default quorum group) and provides five files with right to vote and an OCR and two mirrors of the OCR. With high redundancy, the cluster can survive the loss of two failure groups.

Three of the five discs are used? and two mirrors of OCR? In a Diskgroup single?

Now things goes bad.

As much as I can test and see how Quorum Votedisk four (but not three) of five disks are used and all five counties.

Tags: Database

Application of quorum

Similar Questions

Maybe you are looking for