refactor(CSI-250): do not maintain redundant active mounts from node server after publishing volume

[sergeyberezansky]

TL;DR

Added "sync,async" to mutually exclusive mount options and optimized mount handling in the CSI driver.

What changed?

• Added "sync,async" to the list of mutually exclusive mount options in values.yaml.
• Refactored nfsMounter to remove the mountMap and associated locking mechanism.
• Updated NewMount method to create a new nfsMount object on each call.
• Modified mountWithOptions to merge mutually exclusive options.
• Simplified unmountWithOptions to directly use the new mount object.
• Commented out inactive mount logging and garbage collection functions.
• Updated NodePublishVolume to defer unmounting of the parent filesystem.
• Removed volume unmounting from NodeUnpublishVolume.

How to test?

1. Deploy the CSI driver with the updated chart.
2. Test concurrent mount and unmount operations to ensure stability.
3. Monitor logs for any unexpected behavior or errors related to mount handling.

Why make this change?

This change aims to improve the handling of mount options and simplify the mount management process. By adding "sync" and "async" to the mutually exclusive options, we prevent potential conflicts in mount configurations, which might be caused by translating WekaFS mount options to NFS mount options.
The refactoring of the nfsMounter removes the need for a complex mount map and associated locking, which should lead to better performance and reduced complexity. These changes should result in more reliable and efficient volume mounting and unmounting operations in the CSI driver.

---

[sergeyberezansky]

• chore(deps): update sidecar versions #326
• docs(CSI-254): update official docs link in Helm templates and README #323
• feat(CSI-252): implement kubelet PVC stats #322
• refactor(CSI-250): do not maintain redundant active mounts from node server after publishing volume #320 👈
• feat(CSI-247): implement InterfaceGroup.GetRandomIpAddress() #319
• feat(CSI-249): optimize NFS mounter to use multiple targets #318
• fix(CSI-241): fix unmountWithOptions to use map key rather than options.String() #317
• feat(CSI-245): allow specifying client group for NFS #316
• chore(deps): update packages to latest versions and Go to 1.22.5 #314
• chore(deps): combine chmod with ADD in Dockerfile #313
• feat(CSI-239): moveToTrash does not return error to upper layers #312
• fix(CSI-243): service accounts for CSI plugin assume ImagePullSecret and cause error messages. #311
• feat(CSI-244): match subnets if existing in client rule #315
• fix(CSI-241): conflict in metrics between node and controller #325
• feat(CSI-213): support NFS transport #299
• fix(CSI-241): disregard sync_on_close in mountmap per FS #310
• feat(CSI-253): support custom CA certificate in API secret #324
• dev

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @sergeyberezansky and the rest of your teammates on Graphite

[graphite-app]

Graphite Automations

"Request reviewers once CI passes" took an action on this PR • (09/05/24)

1 reviewer was added to this PR based on Sergey Berezansky's automation.

[sergeyberezansky]

Merge activity

• Sep 12, 6:10 AM EDT: @sergeyberezansky started a stack merge that includes this pull request via Graphite.
• Sep 12, 7:43 AM EDT: Graphite couldn't merge this pull request because a downstack PR feat(CSI-245): allow specifying client group for NFS #316 failed to merge.
• Sep 12, 8:43 AM EDT: @sergeyberezansky started a stack merge that includes this pull request via Graphite.
• Sep 12, 8:45 AM EDT: The Graphite merge of this pull request was cancelled.