feat(upgrade): support engine suspension and target switchover

[derekbit]

Which issue(s) this PR fixes:

Issue longhorn/longhorn#6001

What this PR does / why we need it:

Special notes for your reviewer:

Additional documentation or context

[codecov]

Codecov Report

Attention: Patch coverage is 23.12775% with 349 lines in your changes missing coverage. Please review.

Project coverage is 23.95%. Comparing base (35e00d1) to head (e236459).

Files	Patch %	Lines
app/cmd/dmsetup/dmsetup.go	0.00%	184 Missing ⚠️
pkg/nvme/initiator.go	32.66%	93 Missing and 8 partials ⚠️
app/cmd/nvmecli/nvmecli.go	0.00%	30 Missing ⚠️
pkg/util/device.go	37.50%	14 Missing and 1 partial ⚠️
pkg/nvme/nvmecli.go	20.00%	12 Missing ⚠️
pkg/util/dmsetup.go	89.58%	3 Missing and 2 partials ⚠️
pkg/nvme/nvme.go	33.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #94      +/-   ##
==========================================
+ Coverage   22.78%   23.95%   +1.17%     
==========================================
  Files          33       34       +1     
  Lines        4240     4621     +381     
==========================================
+ Hits          966     1107     +141     
- Misses       3115     3340     +225     
- Partials      159      174      +15     

Flag	Coverage Δ
unittests	23.95% <23.12%> (+1.17%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[DamiaSan]

I see you have implemented the flush command over nvme, but actually lvol bdev doesn't support that. I had implemented it, but this development caused a memory leak in the set external parent functionality, which we use in online rebuilding. I understood that we didn't need the flush over nvme and that we only used it at device mapper layer, so I discarded that commit in the new SPDK branch longhorn-v24.05. I thought to restart working over it, to remove the leak, after 1.7.0.

[derekbit]

in online rebuilding. I understood that we didn't need the flush over nvme and that we only used it at device mapper layer, so I discarded that commit in the new SPDK branch longhorn-v24.05. I thought to restart working over it, to remove the leak, after 1.7.0.

Got it. I think the flush subcommand is fine because I don't use it in the suspension and switchover functionalities. The subcommand is just for making the nvme command more complete. WDYT?

[DamiaSan]

in online rebuilding. I understood that we didn't need the flush over nvme and that we only used it at device mapper layer, so I discarded that commit in the new SPDK branch longhorn-v24.05. I thought to restart working over it, to remove the leak, after 1.7.0.

Got it. I think the flush subcommand is fine because I didn't use in the suspension and switchover functionalities. The subcommand is just for making the nvme command more complete. WDYT?

Ok, this is fine for now. In the future I will work again on the flush implementation so we will have it again.

[DamiaSan]

Just a couple of questions, but globally it LGTM

[derekbit]

@DamiaSan I appended an extra commit for resolving build environment error. Can you review it again? Thank you.

[derekbit]

@DamiaSan Can you help review the latest commit? Thank you.