Mount Options

mount.flexfs start accepts a large set of options. The authoritative list — every flag with its type, default, and one-line description — lives in the mount.flexfs CLI reference. Flags marked internal there are hidden from --help output and are intended for advanced tuning or debugging. This page groups the options by purpose and adds the guidance that does not fit in a reference row.

POSIX and access control

These control POSIX semantics and access enforcement: --acl and --xattr; the mount-behavior flags --noAtime, --noExec, --nonEmpty, and --noSUID; the squashing flags --allSquash, --anonGID, --anonUID, and --rootSquash; and --umask. See the CLI reference for the full list with defaults.

Note

--acl needs kernel POSIX ACL support (Linux 4.9+) to perform well. On current kernels it keeps the kernel’s metadata caches, but reading the ACLs of many files, as ls -l does on a large directory, costs an extra request per file (see POSIX ACL mode). On older kernels, and whenever --rootSquash or --allSquash is set, the mount client checks every operation itself — correct, but with more metadata round trips. In production, set the volume-level acl flag instead of --acl, so that no mount of the volume ignores its ACLs. See Access Control.

Note

--rootSquash, --allSquash, --anonUID and --anonGID may only be passed by root; an unprivileged mount that sets any of them is refused. A volume or token that mandates squashing additionally discards all four local flags, since a mount whose root is squashed chooses neither what it is squashed to nor how widely. Use the matching volume flags to squash mounts whose users you do not control — see Access Control.

Many access control flags (acl, allSquash, noAtime, noExec, nonEmpty, noSUID, rootSquash, xattr) can also be set at the volume level via configure.flexfs. Volume-level settings are combined with mount-level flags — if either side enables a flag, it takes effect.

Not every volume token flag is a mount option. A token carrying admin grants reporting access to the volume’s metadata and changes nothing about the mount, so such a token mounts exactly as one without it — see Volume token flags.

The squash flags are the one exception to that combining rule. A volume or token that mandates squashing replaces the mount’s squash settings rather than being combined with them, so a local --allSquash does not take effect against a volume that mandates only rootsquash — see the paragraph above and Access Control.

Caching

The on-disk cache (--diskFolder, --diskQuota, --diskMaxBlockSize, --diskWriteback, --diskSync) is the part of the caching stack worth tuning; the in-memory block cache and buffer pool are sized automatically and rarely need attention. See the CLI reference for defaults, and the performance tuning guide for when to override them.

Tip

The default --diskFolder path is within the user’s home directory (~/.flexfs/mount/cache/<name>/<pid>). For better performance, consider pointing this to a path on a local SSD. The <name> placeholder is replaced with the volume name and <pid> with the mount process ID.

Caution

A custom --diskFolder should include the <pid> placeholder so each mount process gets its own folder. At startup the mount client removes the flexFS cache files left in its folder; other files in the folder are left alone. A running mount locks its folder, so if two mount processes (or a mount and its auto-update successor) are given the same fixed folder, the second one uses a separate folder beside it, named in the mount log.

Networking

--sse requests S3 server-side encryption for the mount’s block writes. See the CLI reference.

Credentials and auth

--credsFile selects the credentials file (its <name> placeholder is replaced with the volume name). See the CLI reference and the credentials file format.

Update control

--noRemount, --noUpdate, and --staging govern the auto-update mechanism. See the CLI reference and Auto-Update.

Logging and diagnostics

--verbose, --logFile, --metrics / --metricsPort, and --reportErrors are the logging controls. See the CLI reference and the logging guide.

Time-travel

--atTime mounts the filesystem at a point in time (RFC 3339, implies --ro). See Time-Travel Mounting.

Passing options via fstab

Any mount.flexfs start flag can be passed as an fstab mount option by removing the -- prefix. Boolean flags are set by name alone; name=false is accepted too, though every boolean here already defaults to false. For example:

flexfs:<volume-name>  /mnt/data  flexfs  _netdev,nofail,nosuid,x-systemd.mount-timeout=660,ro,acl,verbose  0 0

See fstab Integration for details.