Atime issue moving from sshfs to rclone

STOP and READ USE THIS TEMPLATE NO EXCEPTIONS

What is the problem you are having with rclone?

I am using proxmox backup server to create backups. To that end I am using sshfs to mount a remote folder locally.

It works with sshfs, but is very slow. I am hoping to get a performance boost from using rclone.

This is my fstab entry for sshfs

sshfs#user@user.rsync.net:pbs /pbs-rsyncnet fuse sync_readdir,dir_cache=no,no_readahead,sshfs_sync,max_conns=10,_netdev,uid=34,gid=34,ServerAliveInterval=15,ServerAliveCountMax=3,IdentityFile=/root/.ssh/id_rsync_net,delay_connect,reconnect,allow_other 0 0

With this I can start proxmox backup server’s garbage collection.

When I try the same with the rclone command below I am getting the atime error below.

After reading the documentation I unfortunately still don’t know how to adapt my rclone-fstab entry so that it does “the same” as the sshfs-fstab entry.

Any help is much appreciated.

Run the command 'rclone version' and share the full output of the command.

rclone v1.70.3

  • os/version: debian 12.12 (64 bit)
  • os/kernel: 6.2.16-15-pve (x86_64)
  • os/type: linux
  • os/arch: amd64
  • go/version: go1.24.4
  • go/linking: static
  • go/tags: none

Which cloud storage system are you using? (eg Google Drive)

rsync.net (sftp)

The command you were trying to run (eg rclone copy /tmp remote:tmp)

rsync.net:pbs /pbs-rsyncnet rclone rw,nofail,_netdev,x-systemd.automount,args2env,uid=34,gid=34,vfs_cache_mode=off,buffer_size=64M,multi_thread_streams=4,allow_other,config=/root/.config/rclone/rclone.conf,cache_dir=/var/cache/rclone 0 0

Please run '' and share the full output. If you get command not found, please make sure to update rclone.

rclone config redacted
[rsync.net]
type = sftp
host = XXX
user = XXX
key_file = /root/.ssh/id_rsync_net
shell_type = unix
md5sum_command = md5 -r
sha1sum_command = sha1 -r
### Double check the config for sensitive info before posting publicly

A log from the command that you were trying to run with the -vv flag


2025-09-25T09:13:53+02:00: starting garbage collection on store rsync-net
2025-09-25T09:14:08+02:00: Chunk metadata was not correctly updated during access time safety check:
2025-09-25T09:14:08+02:00: Timestamps before update: accessed Some(SystemTime { tv_sec: 1758783065, tv_nsec: 0 }), modified Some(SystemTime { tv_sec: 1758783065, tv_nsec: 0 }), created None
2025-09-25T09:14:08+02:00: Timestamps after update: accessed Some(SystemTime { tv_sec: 1758783065, tv_nsec: 0 }), modified Some(SystemTime { tv_sec: 1758783065, tv_nsec: 0 }), created None
2025-09-25T09:14:08+02:00: TASK ERROR: atime safety check failed: access time safety check using pre-existing chunk failed, aborting GC!

IMO do not waste your time. PBS is designed to operate on local, ideally SSD disk. Any remotely mounted fs will be painfully slow. No way around it.

What you could try is to use S3 as your datastore location. It has been introduced in the latest PBS release. And here rclone could help with “rclone serve s3” - exposing your remote storage over S3 API

You are probably right, but I don’t have the infra to run everything locally atm and I also need an offsite backup anyway. I am fine with long waiting times. Is there a way to get working atime going?

Interesting note about S3, thank you. I’ll have to check if that is PBS4 or 3(which I am still running).

welcome to the forum,

fwiw, should not trust rclone serve s3 for critical backups.


PBS has added support for S3, but not yet recommend for production.


i have been using rsync.net for 7+ years. something i can trust with critical backups, ZFS snapshots, rsync, restic, and great, unlimited tech support phone calls in case of a disaster.

Good to know, thanks!

I am very happy with rsync.net, I just can’t figure out how to configure rclone the same way as sshfs

this is the remote that i use on windows. i also have a remote for linux, the only change is the location of the keyfile.

[rsync]
type = sftp
host = redacted.rsync.net
user = redacted
key_file = d:\data\c\combined\rsync\keys\profile2\id_ed25519
md5sum_command = md5 -r
sha1sum_command = sha1 -r
shell_type = unix

That seems to be the only difference between our configs. I am still getting the same error though. I guess it might not be possible and I need to find some other way to send my backups to rsync.

no idea about this flag, but it enable atime
--local-time-type mtime|atime|btime|ctime Set what kind of time is returned (default mtime)


Rclone mount does not support atime only mtime and I guess this is the fundamental cause of the problem.

PBS seems to use atime to detect if its garbage collection (GC) is going to work. I don't know if you can disable that or not.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.