About half a year ago we upgraded Zabbix 4 to 6 and upgraded our agents (v1) from 4.4 to Agent2 v6.2 and later v6.4, now 6.4.7.

Since the upgrade from Agent 4.x to Agent2 6.2 all monitored CIFS mounts on all Linux hosts started getting intermittent zero values for all vfs.fs.size items on those mounts.

This only seems to happen on CIFS shares, not on NFS shares or local filesystems.

Steps to reproduce:

1. Ensure a CIFS network share is mounted on Linux host
2. Monitor this share using Agent2 vfs.fs.size

Result:

At irregular intervals the vfs.fs.size items return 0 instead of the real size (used, total, pused doesn't matter, all return 0 at intervals)

See screenshot: the first half of the graphs are monitored by Agent 4.x, then the data gap was the server migration window. And after that Agent 6.2 (later 6.4) was installed on the hosts which intermittent started returning incorrect zero values. (causing flapping of several triggers)

All host (mix of CentOS 7, Oracle Linux 7 and SLES 15 SP3 to SP5 servers) monitoring a CIFS mount have similar graphs.

Expected:
Constant correct values for vfs.fs.size items like it was on Agent v4.x