Revert code added to maintain cephfs mount cache on nodes

[ShyamsundarR]

As functionality added by PR #282 for CephFS fuse mounts does not work as intended, from a code maintainability standpoint we should revert this commit or remove the added code.

[jianglingxia]

if the #282 PR reverted ,the ceph csi driver daemonset plugin existed,caused the pod mount path transport endpoint is not connected error problem can resolved? @ShyamsundarR
thanks

[Madhu-1]

it won't resolve any issue. this code is not working as expected am working on removing the code

[jianglingxia]

@Madhu-1 really thanks, i want to know that the transport endpoint is not connected error ,the problem is cephfs-fuse client reason? not csi driver reason is right?

[Madhu-1]

@jianglingxia see #703 (comment) for the more detailed explanation

[Madhu-1]

closing this one as #825 is merged

[huaizong]

@ShyamsundarR @jianglingxia @Madhu-1 I feel sorry for patch not help, but as PR #282 describe, mount cache patch only work if Pod can restart when csi container exit, it may trigger by health probe check or other similar technology.

If Pod not check mount path, mount cache can not help in thhe scenarios.

NoGoal

Pod should exit and restart when csi plugin pod exit and mount point lost. if pod not exit will get error of transport endpoint is not connected.