Replacing the existing tiebreaker host by a new one

You can replace a dedicated tiebreaker host with another one. During the period where the cluster is without the tiebreaker host, it is expected that alerts to be raised concerning lack of file system tiebreaker group and the cluster having even number of hosts. When the new tiebreaker host is added in and the file system tiebreaker disks be restored, the alerts are cleared automatically.

Before you begin

The instructions that are provided below, use figure 1 as illustrations and have the following assumptions:

All nodes in the cluster are up
All replicated file systems and their respective disks are in "UP" state.
The tiebreaker host is TB (not M5) and all file system tiebreaker disks are local to host TB.
All commands that are provided can be run on any online host in the cluster except for the TB host.
The host name of the new TB host is "TBNew" and it does not have access to redundancy groups 1 and 2.

Figure 1. Multi-site IBM Spectrum Scale replication topology

Procedure

Remove all file system tiebreaker disks owned by host TB first.
```
db2cluster -cfs -remove -filesystem db2sd_1 -disk hdisk7
db2cluster -cfs -remove -filesystem db2datafs -disk hdisk8
db2cluster -cfs -remove -filesystem db2logfs -disk hdisk9
```
If host TB is offline, IBM Spectrum Scale cannot invalidate the disks so that it can be reprovisioned by IBM Spectrum Scale in the future without manual intervention. No further action is required if those disks are not intended to be reused for IBM Spectrum Scale. Otherwise, the first 4K on disk needs to be manually cleared before it can be reused by IBM Spectrum Scale.
Stop host TB in IBM Spectrum Scale and RSCT clusters and remove it from both clusters.
```
db2cluster -remove -host TB
```
Ensure the new tiebreaker host, TBNew, is ready and has the correct Db2® pureScale® installed before proceeding.
Add TBNew to the RSCT and IBM Spectrum Scale clusters.
```
db2cluster -add -host TBNew -no_san_access
```

Add the file system tiebreaker disks to each existing file system. For illustration purpose, this example uses the same disk names removed in step 1.

db2cluster -cfs -add -filesystem db2sd_1 -fstiebreaker hdisk7 -host TBNew
db2cluster -cfs -add -filesystem db2datafs -fstiebreaker hdisk8 -host TBNew
db2cluster -cfs -add -filesystem db2logfs -fstiebreaker hdisk9 -host TBNew

Verify that the disks were added to the appropriate redundancy groups.

db2cluster -cfs -list -filesystem db2sd_1
db2cluster -cfs -list -filesystem db2datafs
db2cluster -cfs -list -filesystem db2logfs