To change the disk configuration:

• If the GUI is accessible, the best is to  add a new disk and use this documented procedure to configure Discovery to use it. See the section "To move data to a new disk". This is the preferred option.
• If the GUI is not accessible, follow this procedure: Discovery: The appliance has shut down and cannot be restarted because a disk is full. This is typically needed when the problem below happens (and the only solution is to add more storage).

There are 2 options to modify the disk configuration without using the GUI:

1. In the case of a virtual appliance, increase the size of the virtual disk hosting the datastore partition and resize it using this documented procedure.
2. add a new disk and use tw_disk_utils to move some data into it as explained in the steps below.

Here is what tw_disk_utils can or can't do:

- The tw_disk_utils utility does not support LVM. 

- The tw_disk_utils utility can only move one set of Discovery data to a given disk. The entire new disk will be used. The data components that can be moved are:

• datastore_data
• datastore_logs
• debug_logs
• record_pool
• backup_data

- Moving a second set of Discovery data to a given disk cannot be done with tw_disk_utils, and is not recommended.
- The tw_disk_utils utility can move more than one set of Discovery data to different disks in the same operation. For example, to move the datastore data to /dev/sdc and the datastore transaction logs to /dev/sdd:

$ tw_disk_utils --move datastore_data:/dev/sdc --move datastore_logs:/dev/sdd

- The tw_disk_utils utility can create a swap partition on the same disk as part of a move operation. For example, suppose the appliance has an existing datastore partition (/mnt/addm/db_data) on /dev/sdc, and that disk also had a swap partition. The datastore has outgrown the current partition and needs to be moved to a new disk. As part of the move operation, it is recommended to also create a new swap partition on the destination disk. If this is not done in one operation, it won't be possible to add a swap partition to the new disk later.
- To use tw_disk_utils to create *only* a swap partition on a new disk, see the following article: Discovery: How to create a new swap partition for the Discovery appliance using tw_disk_utils, and how to remove an unused swap partition

For example: here is the procedure to move the datastore to a new, larger disk, and to create new swap on the new disk:

1) Stop all the Discovery services, then shut down the VM. This can be done from the command line ('sudo shutdown -h now') or simply power off the VM from the VMWare console

2) Use the VMWare console to allocate a new disk of sufficient size for the datastore, plus growth. Do not format or partition this disk.

3) Power on the VM after the disk is added.
 
4) From the command line, login as user tideway and stop the services:​  

$ sudo /sbin/service tideway stop (for RedHat/CentOS 6)
$ tw_service_control --stop (for CentOS 7)

5) Run fdisk ('sudo fdisk -l') to determine the device of the new disk (for example, /dev/sdc). Make a note of this name. 
If new disk does not appear in the "sudo fdisk -l" results, it may be necessary to reboot the appliance and then re-check for the new disk.

6) If the datastore currently resides on its own partition (for example /mnt/addm/db_data on /dev/sdc), do the following to determine if that disk also has a swap partition:
Run "swapon -s". If the output looks like this:

Filename                                Type            Size    Used    Priority
/dev/sdc2                               partition       8388604 2581100 -1

Then it means an 8 gb swap partition exists on /dev/sdc, and a similar partition should be created on the new disk. See below for further steps in this case.

7) Change the working directory to /usr/tideway:

$ cd /usr/tideway

8) Move the datastore and/or create new swap to the new disk with the following commands:

$ screen

The screen command above allows to restore a terminal session if the connection is dropped.

To move the datastore from its default allocation (/dev/sda) to a new, separate datastore partition, the command could look like this:

$ tw_disk_utils --move datastore_data:/dev/sdc   

Replace /dev/sdc with the correct name of the new disk noted in step 5 above. This command will format and partition the new disk and move the datastore to it. 

Note: if moving the datastore transaction logs, use "datastore_logs" instead of "datastore_data" in the command above.

To move from an existing separate datastore partition and also create an 8gb swap partition, the command could look like this:

$ tw_disk_utils --move datastore_data:/dev/sdc  --new-swap 8000:/dev/sdc

Replace /dev/sdc with the correct name of the new disk noted in step 5 above.

9) After the tw_disk_utils is finished, it will say "Completed". At that point, tideway services can be restarted, and the UI should be available.

Use tw_disk_utils --help  to see help for the other features.
 

Notes:
- If the terminal session is lost while tw_disk_utils is running, it can be reattached with "screen -r".
- If "screen -r" does not restore the session (or screen was not run), then the log can be followed like this:

$ tail -f /usr/tideway/log/tw_disk_utils.log

- Some errors are normal and can be ignored, such as the one below.  If provided with the option to proceed in spite of errors, do so.

[tideway@appliance ~]$ tw_disk_utils --move datastore_data:/dev/sdc --new-swap 8000:/dev/sdc
Fri Apr 17 13:57:53 2020 : tw_disk_utils started.
Retrieving the system configuration
ERROR: sudo parted print command error: Error: /dev/sdc: unrecognised disk label
Validating the system configuration
Checking /etc/fstab: all entries must be valid
Verifying that there is enough space in the new disks
Validating the move requests
Estimating the overall process duration
WARNING: tw_disk_utils will shut down BMC Discovery services
The disks that will host the new partitions will be completely
formatted (all data stored in those disks will be deleted)
INFO: rough estimated duration: 9 hours
Do you wish to proceed? (y/n) y

Please also see the following video "How to resolve disk space problems in BMC Discovery".