[ADBDEV-6847] Implement cluster validation possibility

[bimboterminator1]

Implement cluster validation possibility

This is the first commit for building an MVP for new rebalance utility -
gprebalance. This utility is intended to be used for the situation, when after
cluster resize (after expand, shrink) is in unbalanced state. Balanced state
is defined very simple: if number of segments per host is equal across all the
hosts, then cluster is balanced. There are a lot of other aspects for proper
implementation of optimal rebalance algorithm, which will be implemented in
the next patches.

This patch adds the skeleton of future utility, providing initial validation
of rebalance possibility. It includes checks, that validate some basic aspects:
whether segments can be distributed uniformly and can target mirroring strategy
be achieved. Decided to provide validation through separate classes, which is
different approach from gpexpand utility. Also, some unit tests have been added.
Validation of available disk space is not implemented since cannot be achieved at
this initial validation step

---

Unit tests can be run from gpMgmt/bin by running rule make unitdevel

[KnightMurloc]

I think it's worth adding the ability to run a mirrorless rebalance in silent mode. As I understand it, in the current version we require interactive confirmation from the user. It may be worth adding the -y (answer yes to everything) or -a (auto) option, or giving the option to specify mirrorless as mirror-mode.

[KnightMurloc]

I discovered the following situation. If one of the segments is down, gprebalance may mistakenly consider the cluster to be balanced. Example:
There are 3 primary segments on the sdw1 host and 1 primary segment on sdw2. If one of the segments on sdw1 falls, then its role will be performed by a mirror on the sdw2 host, and two primary segments will be obtained on each host. In this case, gprebalance will consider the cluster to be balanced. Is this the expected behavior?

If you specify an incorrect argument, the utility will output several stack traces, which does not look good.

gpadmin@cdw:~/gpdb_src/gpMgmt/bin$ ./gprebalance asdasdasdas
20250121:08:06:13:010313 gprebalance:cdw:gpadmin-[ERROR]:-Unknown argument asdasdasdas
Traceback (most recent call last):
  File "/home/gpadmin/gpdb_src/gpMgmt/bin/./gprebalance", line 289, in main
    options, args = validate_options(options, args, parser)
  File "/home/gpadmin/gpdb_src/gpMgmt/bin/./gprebalance", line 104, in validate_options
    parser.exit(1)
  File "/usr/lib/python3.10/optparse.py", line 1559, in exit
    sys.exit(status)
SystemExit: 1

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/gpadmin/gpdb_src/gpMgmt/bin/./gprebalance", line 344, in <module>
    main(options, args, parser)
  File "/home/gpadmin/gpdb_src/gpMgmt/bin/./gprebalance", line 339, in main
    remove_pid_file(options.coordinator_data_directory)
AttributeError: 'Values' object has no attribute 'coordinator_data_directory'

[KnightMurloc]

In the current version, the indentation on the second line will be included in the final string. The following changes help to avoid this.

Suggested change

	if not gpreb.options.silent and not ask_yesno('', ''' %s Are you sure you want
	to continue with this gprebalance session?''' % str(e), "N"):
	if not gpreb.options.silent and not ask_yesno('', ("%s Are you sure you want"
	"to continue with this gprebalance session?") % str(e), "N"):

[KnightMurloc]

This check is done twice. Do we really need to do this twice? If so, can we make it a separate function?
Also according to the specification we should offer the user to use gpresize, gpexpand or gpshrink to bring the cluster to the desired state.

[KnightMurloc]

why not --target-hosts as in the specification?

[KnightMurloc]

Why not in the current directory? As gpexpand does, for example. It may also be better to print the full path to the file, not just the directory.