Month long ZFS scrubs

TrueNAS General
1

What do we do when we have like 500TB+ and it wants to do month long scrubs?

2

Scrubs should run in parallel, more disks=faster combined speed.

Just because a 6TB pool takes 6 hours to scrub doesn’t mean a 500TB pool will take 500.

If your setup actually takes a month to scrub something is SERIOUSLY wrong either with the drives or the Controller.

1 Like
3

Shouldn’t take a month… Here’s one of my pools, filled to 94.2% capacity – way further than recommended for optimal performance:

Usable Capacity: 557.72 TiB
Last Scrub Date: 2025-03-24 03:23:22
Last Scan Duration: 2 days 2 hours 23 minutes 17 seconds
1 Like
4

Or your CPU is mad old, or you’re running a highly inadvisable drive layout like a 1 vdev 48 drive RAIDZ.

@truenasman28 Is this ia theoritical question, or are you actually experiencing this?

2 Likes
5

Or the pool uses dedup despite RAM being way too low for that. (A 0.5 PB pool would require a few TB of RAM for the DDT…)

@truenasman28 Is it time to actual completion or the (very, very wrong and way off) estimate that is displayed upon initiating a scrub?

1 Like
6

@truenasman28 - Please post the following information, with the command outputs in CODE tags:

  • TrueNAS version
  • Hardware, including disk maker & model
  • How each of the data pool disks is wired to the server, (aka motherboard SATA port?)
  • Output of zpool status and zpool list -v

As others have mentioned, some types of hardware or configurations of the hardware are not suitable. For example, SMR disks have been shown repeatedly not suitable for use with ZFS. And very wide RAID-Zx vDevs have also been shown to become slower as they are used, (fragmented).

7

For the stat collectors, my 2 pool, 9 drive, 54tb array (1 pool at 48% full, the other at 30%) takes exactly 24 hours to scrub. 64mb ddr4 ram, some 2124 Xeon chip. Fast enough for me.