Adds NFS support

[claudiubelu]

Description

Adds support for enabling NFS clusters in microceph.

Adds the following commands:

• microceph enable nfs --cluster-id <cluster-id> [--bind-address <ip-address>] [--port <port-num>] [--v4-min-version 0/1/2] [--target <server>] [--wait <bool>]
• microceph disable nfs --cluster-id <cluster-id> [--target <server>]

Adds the services/nfs API endpoint (PUT / DELETE).

Adds the following tables:

• service_groups: used to track microceph service clusters.
• grouped_services: used to track clustered services running on a particular server.

Added dependencies in the snap:

• nfs-ganesha
• nfs-ganesha-ceph
• nfs-ganesha-rados-grace

Type of change

• New feature (non-breaking change which adds functionality)

How has this been tested?

Note

All functional changes should accompany corresponding tests (unit tests, functional tests, etc.).

Please describe the addition/modification of tests done to verify this change. Also list any
relevant details for your test configuration.

Contributor checklist

Please check that you have:

• self-reviewed the code in this PR
• added code comments, particularly in less straightforward areas
• checked and added or updated relevant documentation
• checked and added or updated relevant release notes
• added tests to verify effectiveness of this change

[UtkarshBhatthere]

first round of reviews: some comments and questions. Thanks @claudiubelu!

[UtkarshBhatthere]

1. Not providing a target name does not default to the local host.
2. The service name should default to nfs rather than nfs-ganesha (as we enable nfs not nfs-ganesha).

root@microceph1:~# microceph enable nfs --cluster-id zuzu
Error: failed placing service nfs: Failed to get cluster member for request target name "0.0.0.0:2049": CoreClusterMember not found
root@microceph1:~# microceph enable nfs --cluster-id zuzu --target microceph1
root@microceph1:~# microceph status
MicroCeph deployment summary:
- microceph1 (10.17.213.183)
  Services: mds, mgr, mon, nfs-ganesha, osd
  Disks: 1
- microceph2 (10.17.213.150)
  Services: mds, mgr, mon, osd
  Disks: 1
- microceph3 (10.17.213.215)
  Services: mds, mgr, mon, osd
  Disks: 1

[gboutry]

@UtkarshBhatthere Any updates? Can you review please?

[UtkarshBhatthere]

@gboutry I will do a detailed review today

[UtkarshBhatthere]

Enabling a second daemon (on a multinode microceph cluster fails)

@claudiubelu

Node 1

root@secondary:~# microceph enable nfs --cluster-id manila
root@secondary:~#

Node 2

# enablement on second cluster fails
root@primary:~# microceph enable nfs --cluster-id manila
Error: failed placing service nfs: failed to add DB record for nfs: conflicting service group configurations
root@primary:~#

Database Dump

root@primary:~# microceph cluster sql "select * from grouped_services"
+----+------------------+-----------+---------------------------------------------+
| id | service_group_id | member_id |                    info                     |
+----+------------------+-----------+---------------------------------------------+
| 1  | 1                | 2         | {"bind_address":"0.0.0.0","bind_port":2049} |
+----+------------------+-----------+---------------------------------------------+
root@primary:~# microceph cluster sql "select * from service_groups"
+----+---------+----------+----------------------+
| id | service | group_id |        config        |
+----+---------+----------+----------------------+
| 1  | nfs     | manila   | {"v4_min_version":1} |
+----+---------+----------+----------------------+
root@primary:~#

[UtkarshBhatthere]

root@primary:~# microceph status
MicroCeph deployment summary:
- primary (10.206.113.174)
  Services: mds, mgr, mon, osd
  Disks: 4
- secondary (10.206.113.47)
  Services: mds, mgr, mon, nfs.manila
  Disks: 0
root@primary:~# microceph enable nfs
Error: please provide a cluster ID using the `--cluster-id` flag
root@primary:~# microceph enable nfs --cluster-id manila
root@primary:~# microceph status
MicroCeph deployment summary:
- primary (10.206.113.174)
  Services: mds, mgr, mon, nfs.manila, osd
  Disks: 4
- secondary (10.206.113.47)
  Services: mds, mgr, mon, nfs.manila
  Disks: 0
root@primary:~#

[UtkarshBhatthere]

lgtm. thanks @claudiubelu

[sabaini]

Thanks @claudiubelu ! Added some commentary

[claudiubelu]

Please also consider adding a functional test and documentation.

Added functional tests (single node and multinode). Will add documentation.

A general request, it would be great if we could get more logging of key operations.

That would entail logging of errors, as well as debug logs e.g. for entry and exit of public functions -- especially on the microcephd side

Added more logging, and reverts to EnableNFS.

1. The output of `microceph status` does not report nfs service.

Wasn't sure how you envisioned this to be reported exactly, but I see you've added a commit for that, and now it reports nfs.<cluster_id>. Thanks!

2. There is a mismatch in the underlying service name (`snap.microceph.nfs-ganesha`) and the microceph command name `microceph enable nfs`. In my opinion it should be called `nfs` everywhere accessible to users (CLI commands, snapcraft service name, etc) . Ans should use `nfs-ganesha` in implementation (like actual service name, and expected paths etc)

Done, now the service name is snap.microceph.nfs.

2. There is a mismatch in the underlying service name (`snap.microceph.nfs-ganesha`) and the microceph command name `microceph enable nfs`. In my opinion it should be called `nfs` everywhere accessible to users (CLI commands, snapcraft service name, etc) . Ans should use `nfs-ganesha` in implementation (like actual service name, and expected paths etc)

Enabling a second daemon (on a multinode microceph cluster fails)

Resolved, added another functional test for the multinode scenario. It passes.

[sabaini]

Thanks @claudiubelu!
We will need docs for this too but from a functionality perspective this looks good.

[claudiubelu]

Thanks @claudiubelu! We will need docs for this too but from a functionality perspective this looks good.

Added the documentation update here #583, so we can discuss any changes needed to it there, so we can proceed with merging this.

[UtkarshBhatthere]

Thanks for the amazing contribution @claudiubelu 🚀