a.gritsenkoandClaude Opus 5.5
42cfb43e2e
build / build (push) Successful in 1m5s
RepoISO now manages the NAS itself, so nasmon no longer polls it (settings field, dashboard card, menu link and repoiso:* alerts are gone). Drop-in SMB includes, read-only listing and keeping smbd/nfs-server running for other software's shares stay. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
64 lines
3.2 KiB
Markdown
64 lines
3.2 KiB
Markdown
# nasmon
|
||
|
||
Lightweight NAS health monitor: web dashboard on port 8090 with CPU, memory
|
||
(incl. ZFS ARC), temperatures, physical disks with SMART, ZFS pools and datasets;
|
||
periodic SMART/ZFS checks with e-mail alerts.
|
||
|
||
* Web UI: http://<host>:8090
|
||
* Settings: http://<host>:8090/settings (also /etc/nasmon/config.json)
|
||
* State: /var/lib/nasmon/nasmon.db (SMART history, alert log)
|
||
* Service: systemctl status nasmon ; journalctl -u nasmon -f
|
||
* JSON API: /api/status, /api/history?hours=24, /api/disks/<serial>/history?days=30
|
||
|
||
## ZFS datasets from the UI
|
||
|
||
The ZFS page (http://<host>:8090/zfs) can create datasets in a pool — name (nested
|
||
names like media/video are fine), owner uid:gid for the mountpoint, compression,
|
||
dedup, optional quota — and destroy datasets that are empty (no data, no children, no
|
||
snapshots; the check is repeated against the live system before "zfs destroy").
|
||
Each pool card also shows the dedup ratio and can turn dedup on/off on the pool
|
||
root dataset (`zfs set dedup=on|off <pool>`, inherited by child datasets; only new
|
||
writes are affected, and the dedup table needs a lot of RAM).
|
||
|
||
nasmon runs sandboxed (ProtectSystem=strict → own mount namespace), so these
|
||
commands are executed on the host through transient units:
|
||
`systemd-run --wait --pipe --collect zfs create …`. Every action is logged
|
||
(journalctl -u nasmon).
|
||
|
||
nasmon has no authentication: keep port 8090 on the LAN only.
|
||
|
||
## SMB / NFS shares from the UI
|
||
|
||
http://<host>:8090/shares manages Samba shares (guest or per-user, files always
|
||
written as the data owner uid 1000), SMB users (smbpasswd; a login-less Unix
|
||
account is created when needed) and NFS exports (host/IP/CIDR clients, rw/ro,
|
||
squash). A share can point to any existing folder (ZFS dataset, USB disk under
|
||
/mnt or /media, /srv…) except system directories (/etc, /usr, /var, /boot…),
|
||
also after resolving symlinks. Definitions are stored in
|
||
/etc/nasmon/shares.json; nasmon regenerates
|
||
/etc/samba/smb.conf (whole file, header "Managed by nasmon") and
|
||
/etc/exports.d/nasmon.exports, validates with testparm, then reloads smbd /
|
||
exportfs -ra. A rejected config is rolled back to the .bak copy. Daemons are
|
||
enabled while at least one share exists and disabled otherwise. "Allow legacy
|
||
clients" in the SMB settings turns on SMB1 + NTLMv1 (server min protocol = NT1)
|
||
for IPMI/BMC virtual media and other old devices; it is off by default.
|
||
|
||
Alerts: the × on an open alert acknowledges it (hidden and not mailed until the
|
||
problem changes or clears); the × on a resolved alert removes it from the log.
|
||
Resolved alerts older than 7 days are hidden by default, older than 90 days
|
||
deleted.
|
||
|
||
## Shares of other services
|
||
|
||
Other software on the host can publish shares without fighting over the
|
||
config files nasmon owns:
|
||
|
||
* SMB: drop a fragment into /etc/samba/smb.conf.d/<name>.conf. nasmon appends
|
||
"include = <file>" for every *.conf there to the smb.conf it generates.
|
||
* NFS: write /etc/exports.d/<name>.exports.
|
||
|
||
nasmon lists these shares read-only on /shares and keeps smbd / nfs-server
|
||
enabled while any of them exist, even with no nasmon shares (previously it
|
||
disabled the daemons and took such shares down). When SMB is switched off in
|
||
nasmon, only nasmon's own shares are removed.
|