Search Results
6/13/2025, 12:02:20 PM
>>105580613
> zfs pool with multiple datasets
> passed to vm
> smb to identical vm
> Transfers 150-300MB/s for 20 seconds
> Freezes for a 1:30 to 2:00
> Returns to copying
How fast are the drives? SMB and iSCSI both add overhead, SMB much moreso.
It sounds to me like you're completely flooding cache (Either VM RAM, ARC, or Disk) and that data has to get flushed to disk so more data can be transfered. But it shouldn't *freeze*. Weird caching behavior maybe? It could also be a combination of behavior in each layer.
Even on spinning rust I can manage 140mb/s for a few seconds before getting throttled down to 60-80MB/s but that's consistent speed. When I was doing iscsi to a VM and then SMB across VMs (Bridging my Jellyfin and ingest servers) I'd manage 40mb/s with freezes between files.
> zfs pool with multiple datasets
> passed to vm
> smb to identical vm
> Transfers 150-300MB/s for 20 seconds
> Freezes for a 1:30 to 2:00
> Returns to copying
How fast are the drives? SMB and iSCSI both add overhead, SMB much moreso.
It sounds to me like you're completely flooding cache (Either VM RAM, ARC, or Disk) and that data has to get flushed to disk so more data can be transfered. But it shouldn't *freeze*. Weird caching behavior maybe? It could also be a combination of behavior in each layer.
Even on spinning rust I can manage 140mb/s for a few seconds before getting throttled down to 60-80MB/s but that's consistent speed. When I was doing iscsi to a VM and then SMB across VMs (Bridging my Jellyfin and ingest servers) I'd manage 40mb/s with freezes between files.
Page 1