监控项
NAS 与备份
Synology、TrueNAS、Unraid、Proxmox Backup Server 和 Veeam:磁盘、存储池,以及最近一次成功备份的时间。
监控项描述和告警规则文本为英文,与产品一致。页面标签使用您的语言。
Volumes, pools, disk health, temperature, Hyper Backup and Active Backup jobs, Synology Drive and Photos.
示例
内置告警规则 17
- 提示Synology disk SMART warningDSM no longer reports this disk's SMART status as normal.
- 警告Synology disk failedDSM reports the disk as crashed or failed.
- 警告Synology disk bad sectorsThe disk's bad-sector count exceeds the threshold set in DSM.
- 提示Synology disk bad sectors growingNew unreadable sectors appeared in the last 24 hours (delta(dumbmonit_synology_disk_unc_count[24h])).
- 提示Synology SSD wearing outAn SSD has less than 10 % of its rated life left; clears above 12 %.
- 警告Synology volume degradedA volume is degraded or crashed: its redundancy is gone.
再显示 11 条
- 警告Synology storage pool degradedA storage pool or SSD cache is degraded or crashed: a disk failed and the redundancy is gone or rebuilding.
- 提示Synology volume almost fullA volume is 90 % full or more, not counting the space reserved by thick-provisioned iSCSI LUNs; clears below 88 %. A volume filled by a thick LUN is full by design and does not alert.
- 提示Synology thin LUN over-allocationThe thin LUNs of a volume were promised more space than the volume still has free; clears below 95 %.
- 提示Synology thin LUN almost fullA thin LUN has 95 % or more of its size written. A thick LUN is opaque from the NAS: the rules of the monitor that uses it (datastore, filesystem) alert instead.
- 提示Synology temperature highA disk is above 55 °C, or DSM raised its own temperature warning; clears below 52 °C.
- 提示Synology memory highMemory usage of the NAS above 95 %; clears below 90 %.
- 警告Active Backup task failedThe last run of an Active Backup for Business task failed, or backed up only part of its devices. The notification names the task.
- 提示Active Backup too oldNo successful Active Backup for Business run for this task for more than two days. The series does not exist for a task that has never succeeded; "Active Backup task failed" speaks then.
- 信息Active Backup task disabledThe task has no schedule, or its continuous backup is paused: it will not back up anything until someone runs it.
- 提示Active Backup device overdueA device has gone longer without a successful Active Backup for Business run than its own rhythm allows, off-days excluded).
- 警告Active Backup device failingThe last two or more attempts of a device failed; a cancelled run counts as neither.
设置步骤: Create a monitoring account in DSM
In DSM, open Control Panel → User & Group → User → Create and name the account as follows, with a long password used nowhere else.
dumbmonit
Join groups: tick administrators.
Assign shared folder permissions: tick No access on every shared folder.
Assign application permissions: Deny every application except DSM, and Active Backup for Business if you want its tasks read.
Keep two-step verification off for this account (Control Panel → Security → Account).
In DumbMonit, enter the NAS address (HTTPS, port 5001 by default) and this account's user name and password.
注意. The administrators group is required by DSM's storage API, not by DumbMonit: that is why the account gets no shared folder and no other application.
文档
Pool health, the failing disk, scrubs, snapshots, replication and TrueNAS's own alerts.
示例
- TrueNAS SCALE
- TrueNAS Community Edition
您需要准备
- API keyThe key TrueNAS shows once when it is created, as "3-AbCd…".
默认端口: 443
内置告警规则 13
- 警告NAS pool degradedZFS no longer calls a pool healthy. A degraded pool still serves its data, which is exactly why nobody notices until the next disk fails. The monitor page names the disk.
- 提示NAS disk errorsDisks of a pool have reported read, write or checksum errors since the last zpool clear.
- 提示NAS pool almost fullA pool is more than 85 % full. ZFS slows down well before it is full. Clears at 80 %.
- 警告NAS scrub found errorsThe last scrub of a pool found damaged data.
- 信息NAS scrub overdueA pool has not completed a scrub in more than 45 days. Absent after a resilver, which erases the record of the last scrub.
- 警告NAS disk failed its SMART testA SMART self-test in a disk's log failed. A disk never tested has no series.
再显示 7 条
- 提示NAS disk too hotA disk is above 55 °C. Clears at 50 °C.
- 提示NAS dataset near its quotaA dataset has used more than 90 % of its quota: at 100 % its writes fail while the pool still has room. No quota, no series.
- 提示NAS replication failedAn enabled replication task is in error: the copy on the other side is getting older.
- 提示NAS snapshot task failedAn enabled periodic snapshot task is in error.
- 提示NAS snapshots staleAn enabled periodic snapshot task has not run for more than eight days: a task that stops running does not fail, it goes quiet.
- 提示TrueNAS alert raisedTrueNAS itself has raised an alert of level ERROR or above. Its own sentence is on the monitor page.
- 提示NAS service stoppedA service set to start with the NAS is stopped. A service switched off on purpose has no series.
设置步骤: Create an API key for DumbMonit in TrueNAS
Open Credentials → Groups → Add, name the group as follows and give it the Local Administrator privilege.
dumbmonit
Open Credentials → Users → Add, name the account as follows and make dumbmonit its primary group. Leave shell access, sudo and SSH off.
dumbmonit
Open Credentials → Users → API Keys → Add, pick the dumbmonit user, name the key and copy it: TrueNAS shows it only once.
On TrueNAS 24.10, the path is the user menu at the top right → API Keys → Add.
In DumbMonit, enter the NAS address (for example nas.lan) and paste the key. Keep HTTPS.
注意. This key can change anything on the NAS, because TrueNAS offers no read-only key over its REST API: keep it for DumbMonit alone and revoke it if it leaks. TrueNAS ships a self-signed certificate: if the connection is refused, tick "Accept an unverifiable certificate" in the options.
文档
An Unraid server through its GraphQL API: array stopped, disks disabled or with errors, parity checks, cache pool usage, containers that should be running.
示例
- Unraid 7
- Unraid 6.12 with Unraid Connect
您需要准备
- API keyA key with the Viewer role: it reads everything and can change nothing.
默认端口: 80
内置告警规则 9
- 警告Unraid array stoppedThe Unraid array is not started: shares, containers and VMs that live on it are unavailable.
- 警告Unraid disk disabled or missingAn Unraid disk is disabled, missing or invalid: the array runs degraded, and one more failure loses data.
- 提示Unraid disk read errorsAn Unraid disk has reported read errors since the array started: it may be disabled at the next one.
- 提示Unraid disk too hotAn Unraid array or parity disk has stayed above 55 °C for fifteen minutes.
- 提示Unraid parity check found errorsThe last Unraid parity check failed or found sync errors: parity no longer matches the data.
- 提示Unraid parity check overdueThe last Unraid parity check is more than 40 days old: a monthly check was missed.
再显示 3 条
- 提示Unraid cache nearly fullAn Unraid cache pool is more than 90 % full: writes fall back to the slow array, or fail.
- 提示Unraid container stoppedA Docker container set to start with the array is not running.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Create a read-only API key for DumbMonit
On Unraid 6.12 to 7.1, install the Unraid Connect plugin from the Apps tab first. Unraid 7.2 or later has the API built in.
Open Settings → Management Access → API Keys and create a key named as follows, with the Viewer role.
dumbmonit
Copy the key: Unraid shows it only once.
In DumbMonit, enter the Unraid address (for example tower.lan), paste the key as API key and choose https if the web interface is served over HTTPS.
注意. Unraid reports no temperature for a disk that is spun down, and DumbMonit never wakes one: a sleeping disk simply has no temperature until it spins up.
文档
QNAP QTS / QuTS hero
测试版qnap
A QNAP NAS through the CGI pages of its web interface: volumes filling up, disks whose SMART health is no longer good or too hot, processor temperature, firmware updates.
示例
- QNAP TS-453D
- QuTS hero NAS
- QTS 5.x
您需要准备
- User name and passwordThe dumbmonit account created on the NAS. The password is sent base64-encoded to the login page, as the web interface does.
默认端口: 443
内置告警规则 6
- 提示QNAP volume nearly fullA QNAP volume is more than 90 % full: shares, snapshots and apps stop being able to write.
- 提示QNAP disk health not goodThe SMART health of a QNAP disk is no longer good (warning or abnormal): replace it before the RAID loses redundancy.
- 提示QNAP disk too hotA QNAP disk has stayed above 55 °C for fifteen minutes.
- 提示QNAP processor too hotThe QNAP processor has stayed above 90 °C for fifteen minutes.
- 信息QNAP firmware updateQNAP offers a newer QTS or QuTS hero release.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Create a limited account for DumbMonit
Open Control Panel → Privilege → Users → Create → Create a User and create the account as follows, with a long random password. Tick no share and no application privilege.
dumbmonit
Add the account to the administrators group: QTS serves the system status, volumes and disks to that group only.
Optional: under Control Panel → System → Security → Allow/Deny List, allow the web interface only from the DumbMonit server's address.
In DumbMonit, enter the NAS address (for example nas.lan) with that user name and password. Choose http and port 8080 only if HTTPS is off.
注意. QTS has no read-only role for its system status, so the account sits in the administrators group: give it a unique password, and an account without 2-step verification. Over HTTP the password travels base64-encoded, not encrypted: prefer HTTPS.
文档
A Ceph cluster through the REST API of the manager's dashboard: health and its checks, OSDs down or nearly full, placement groups not clean, capacity, pools, monitors out of the quorum.
示例
- Ceph 18 (Reef)
- Ceph 19 (Squid)
- cephadm clusters
您需要准备
- Dashboard user and passwordThe dumbmonit user of the Ceph dashboard, with the read-only role.
默认端口: 8443
内置告警规则 10
- 警告Ceph health errorThe Ceph cluster reports HEALTH_ERR: data may be unavailable or at risk.
- 提示Ceph health warningThe Ceph cluster has reported HEALTH_WARN for 30 minutes: see its health checks.
- 提示Ceph OSD downOne or more Ceph OSDs are down: the cluster rebuilds their data elsewhere and loses redundancy meanwhile.
- 提示Ceph OSD nearly fullA Ceph OSD is more than 85 % full (the default nearfull ratio): writes to it stop at the full ratio.
- 提示Ceph placement groups not cleanSome Ceph placement groups have not been active+clean for 30 minutes: data is degraded, misplaced or inactive.
- 警告Ceph objects unfoundCeph reports unfound objects: no copy of some data can be found.
再显示 4 条
- 提示Ceph capacity highThe Ceph cluster is more than 80 % full: it turns read-only at 95 %, and rebuilding after a failure needs free space.
- 提示Ceph pool nearly fullA Ceph pool is more than 90 % full.
- 警告Ceph monitor out of quorumA Ceph monitor is out of the quorum: one more loss and the cluster stops.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Create a read-only dashboard user for DumbMonit
Check that the dashboard is enabled and note the address of the active manager.
ceph mgr services
Write a long random password to a file, and keep it: DumbMonit needs it.
openssl rand -base64 24 | tr -d '\n' | tee /tmp/dumbmonit.pwd
Create the user with the built-in read-only role.
ceph dashboard ac-user-create dumbmonit -i /tmp/dumbmonit.pwd read-only
Delete the password file.
shred -u /tmp/dumbmonit.pwd
In DumbMonit, enter the address of the active manager (for example ceph.lan) with that user name and password.
注意. A standby manager redirects to the active one: use the address of the active manager. The dashboard ships a self-signed certificate: tick "Accept an unverifiable certificate" unless you installed your own.
文档
A MinIO object store through its Prometheus metrics: nodes and drives offline, capacity, failed heals, replication failures and S3 server errors.
示例
- MinIO single node
- MinIO distributed cluster
您需要准备
- Prometheus tokenThe bearer_token printed by mc admin prometheus generate. It opens the metrics pages only.
无需账户。
默认端口: 9000
内置告警规则 7
- 警告MinIO node offlineA MinIO node is offline: the cluster runs with less parity.
- 提示MinIO drive offlineA MinIO drive is offline: objects are rebuilt from parity until it is back.
- 提示MinIO capacity highThe usable capacity of the MinIO cluster is more than 85 % used.
- 提示MinIO heal failuresMinIO could not heal some objects in its current self-healing run.
- 提示MinIO replication failuresObjects of a MinIO bucket failed to replicate to the remote site.
- 提示MinIO S3 server errorsMore than 5 % of the S3 requests MinIO served in the last 10 minutes failed with a server error (5xx) (100 * rate(dumbmonit_minio_s3_requests_5xx_total[10m]) / rate(dumbmonit_minio_s3_requests_total[10m])).
再显示 1 条
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Generate a Prometheus token for DumbMonit
With the mc client configured for your MinIO (alias myminio), generate the token of the cluster metrics.
mc admin prometheus generate myminio cluster
Copy the value after bearer_token in the output: that is the token.
In DumbMonit, enter the MinIO address (for example minio.lan), paste the token as Prometheus token and choose https if MinIO serves TLS.
注意. The token only opens the metrics pages, but anyone holding it can read them. If MinIO runs with MINIO_PROMETHEUS_AUTH_TYPE=public, the metrics need no token: choose No authentication.
文档
Backup freshness per machine, failed tasks, sync and verify jobs, datastore space.
示例
您需要准备
- API token (recommended)No expiry and no session opened: the right choice for monitoring.
- Username / passwordOpens a two-hour session, renewed automatically. Use it only if tokens are not an option.
默认端口: 8007
内置告警规则 21
- 提示PBS datastore almost fullPBS datastore more than 90% full.
- 提示PBS datastore filling upAt the current rate, PBS estimates the datastore full within seven days.
- 提示PBS datastore not mountedA removable datastore is not mounted: nothing can be written to it.
- 提示PBS backup too oldNo new snapshot for this backup group (one per machine) for more than two days. Edit the threshold for machines backed up weekly; the notification names the group and datastore.
- 警告PBS backup verification failedVerification of the latest snapshot for this machine failed.
- 提示PBS task failedAt least one PBS task failed in the review window.
再显示 15 条
- 提示PBS garbage collection too oldNo successful garbage collection on this datastore for more than eight days.
- 提示PBS garbage collection failedThe last garbage collection on a datastore failed.
- 警告PBS corrupt chunks foundThe garbage collection found unreadable chunks on a datastore: some backups can no longer be restored.
- 提示PBS prune job failedThe last run of a PBS prune job failed: nothing is pruned and the datastore keeps filling up. The notification names the job and datastore.
- 提示PBS verification job failedThe last run of a PBS verification job failed.
- 警告PBS sync job failedThe last run of a PBS sync job failed.
- 警告PBS tape backup failedThe last tape backup run failed: the offline copy is not being made.
- 提示PBS job never ranA job is enabled and scheduled but has never run.
- 警告PBS service downA service Proxmox Backup Server needs is not running; no backup can be taken in while it is down.
- 警告PBS disk SMART failureA disk of the backup server reports SMART health FAILED. The notification reads /dev/sdb.
- 提示PBS SSD worn outAn SSD of the backup server has used more than 90 % of its rated endurance.
- 警告PBS ZFS pool degradedA ZFS pool of the backup server is not ONLINE.
- 提示PBS certificate expiringThe certificate served by the backup server expires in less than three weeks. Needs the certificates option.
- 信息PBS restart pending after upgradeThe package was upgraded but the running daemon is still the old one.
- 信息PBS updates pendingMore than twenty package updates are pending on the backup server.
设置步骤: Create a read-only user and token in Proxmox Backup Server
Open Configuration → Access Control → User Management → Add, enter the user name dumbmonit, choose the realm Proxmox Backup authentication server, then click Add.
dumbmonit@pbs
Open Configuration → Access Control → API Token → Add: user dumbmonit@pbs, token name monitor. Copy the Token ID and the Secret now, the secret is shown only once.
dumbmonit@pbs!monitor
Open Configuration → Access Control → Permissions → Add → User Permission: path /, user dumbmonit@pbs, role Audit.
Add a second permission with Add → API Token Permission: path /, token dumbmonit@pbs!monitor, role Audit. A token has its own permissions, so both are needed.
Optional, to see pull sync jobs and the tape tier: repeat both permissions with role RemoteAudit on path /remote and role TapeAudit on path /tape.
In DumbMonit, enter the server address (for example pbs.lan:8007), the Token ID and the Secret.
注意. Both the user and the token need the permission: given to the token alone, it grants nothing and PBS reports nothing. The server uses a self-signed certificate by default: tick "Accept an unverifiable certificate" in the options if the connection is refused.
文档
Veeam Backup & Replication
veeam
A Veeam backup server through its REST API: jobs whose last run failed, failed sessions of the last 24 hours, repositories filling up, license expiry.
示例
- Veeam Backup & Replication 12
- Veeam Backup & Replication 13
- Veeam Community Edition
您需要准备
- User name and passwordThe account given the Veeam Backup Viewer role, in the form HOST\user or DOMAIN\user.
默认端口: 9419
内置告警规则 6
- 提示Client device staleA device has not connected or not backed up within the staleness window configured for this target.
- 警告Veeam job failedThe last run of an enabled Veeam job failed: what it protects has no new restore point.
- 提示Veeam job ended with a warningThe last run of an enabled Veeam job ended with a warning: some objects may not be protected.
- 提示Veeam repository nearly fullA Veeam backup repository is more than 90 % full: the next jobs will fail for lack of space.
- 提示Veeam license expiringThe Veeam license expires in less than 30 days: jobs stop at the end of the grace period.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Give DumbMonit a Veeam Backup Viewer account
On the backup server, open Computer Management → Local Users and Groups → Users → New User and create the account as follows, with a long random password.
dumbmonit
In the Veeam console, open the main menu > Users and Roles > Security, click Add, enter the account (for example VBR01\dumbmonit) and give it the Veeam Backup Viewer role only.
In DumbMonit, enter the backup server address (for example vbr.lan) and the account as HOST\dumbmonit or DOMAIN\dumbmonit, with its password.
Tick Accept an unverifiable certificate unless you installed your own: the REST API (port 9419) uses a self-signed one.
注意. Veeam keeps some REST API routes for the Backup Administrator role, the license among them on current versions: with the Viewer role DumbMonit shows the license as not readable and never asks for more.
文档
A Kopia server through its REST API: the last snapshot of each source (age, errors, size, files), sources that missed their schedule, and the repository connection.
示例
- Kopia server
- KopiaUI
- Kopia in Docker
您需要准备
- Server user name and passwordThe user name and password the Kopia server was started with.
无需账户。
默认端口: 51515
内置告警规则 4
- 提示Kopia snapshot with errorsThe last snapshot of a Kopia source finished with errors: some files were not saved.
- 提示Kopia snapshot too oldA scheduled Kopia source has no snapshot for more than two days.
- 警告Kopia repository disconnectedThe Kopia server is not connected to its repository: no snapshot can run.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Start the Kopia server with its own user name and password
On the machine that runs the backups, start the Kopia server with a generated certificate and its own user name and password.
kopia server start --address=0.0.0.0:51515 --tls-generate-cert --server-username=dumbmonit --server-password=PASTE-A-LONG-RANDOM-PASSWORD
Check that the repository is connected to that server and that the DumbMonit server can reach port 51515.
In DumbMonit, enter the server address (for example nas.lan) with that user name and password.
Tick Accept an unverifiable certificate: the generated certificate is self-signed. If the server runs with --insecure, choose the http protocol instead.
注意. Kopia's API has no read-only account: the server password also lets its holder start and delete snapshots. DumbMonit only reads, but keep port 51515 on your local network.
文档
A Duplicati server through its API: the age of each task's last successful backup, its last result, duration, size, warnings and errors, and the task running now.
示例
- Duplicati 2.1
- Duplicati in Docker
- Duplicati on a NAS
您需要准备
- Web interface passwordDuplicati has no user accounts: the password of its web interface is all it asks for.
默认端口: 8200
内置告警规则 3
- 警告Duplicati backup failedThe last run of a Duplicati backup failed: what it protects has no new version.
- 提示Duplicati backup too oldA scheduled Duplicati backup has not succeeded for more than two days.
- 警告Monitor unreachableNo measurement received for more than three minutes. bottomk by (target) (1, time() - tlast_over_time(dumbmonit_up[7d]))
设置步骤: Let DumbMonit read the Duplicati API
In the Duplicati web interface, open Settings and set a long random password for the interface: Duplicati 2.1 requires one.
If the server only listens on the machine itself, start it with these options so that DumbMonit can reach it.
--webservice-interface=any --webservice-allowed-hostnames=*
In DumbMonit, enter the Duplicati address (for example nas.lan) and paste that password.
注意. Duplicati 2.1 or later is required: version 2.0 signs in through a different mechanism that DumbMonit does not use. The password gives full access, so keep port 8200 on your local network.
文档