← All break/fix scenarios in Proxmox VE
Replication job is stuck and not making progress
Reported symptoms
- ●pvesr status shows the job with last_sync time = hours ago
- ●Replication task in the GUI shows "active" but no progress
- ●ZFS send/receive is not moving data (zfs send on source, zfs receive on target)
- ●Network between source and target is slow or saturated
Evidence
- · pvesr status (look for last_sync age)
- · ps aux | grep "zfs send\|zfs receive" on both nodes
- · mtr -n <target-node> shows latency or loss
- · zfs list -t snapshot on source pool (snapshots accumulating without being sent)
Diagnosis and resolutionclick to reveal
Root cause
Common causes: - Network bandwidth saturation between source and target - A previous replication left a partial stream (ZFS send interrupted) - The target ZFS pool is full or has no space for new snapshots - The job schedule is too aggressive (snapshot creation too frequent)
Remediation
1. Check the job state: `pvesr status` 2. Find the running replication process: `ps aux | grep pvesr` (or `zfs send` / `zfs receive`) 3. Check target pool capacity: `zfs list <target-pool>` If full, prune old snapshots or expand the pool 4. Check network bandwidth: `iftop -i <replication-iface>` If saturated, throttle: `--rate 50000` (KB/s) on the job 5. If a previous stream is hung: `pvesr stop <job-id>` Manually clean up partial snapshots: `zfs list -t snapshot | grep <target-pool>` `zfs destroy <hung-snapshot>` (carefully) Re-enable the job: `pvesr enable <job-id>` 6. Run a one-off sync to confirm health: `pvesr schedule-now`
Verification
- `pvesr status` shows last_sync within the schedule window - ZFS snapshots on source are being pruned (kept in sync with target) - Network bandwidth is reasonable (< 80% of link capacity) - No "stuck" processes for replication
Prevention
- Schedule replication during off-peak hours - Use bandwidth limits (--rate) to avoid saturating the link - Monitor replication lag with an alert - Plan target storage capacity for the replication rate - Test replication periodically to a separate target