Redundant SLOG device?

[Exhorder]

Hi,

I've got a question about SLOG. I'm not sure if it has to be redundant. What I've read so far says, that it does not need to be redundant because if device fails, the ZIL just goes to the pool where it would be without SLOG anyways.
OK, but I never had a completely failing SSD so far. If I had a failing SSD it just resulted in corrupted file systems (ext4 or XFS) because apparantly corrupted data got written to the disk. If this happens to my SLOG, what is the result? Is data on SLOG also checked before it goes to the pool? Or would only a redundant SLOG help here? Or would a redundant SLOG help here anyway?

Thanks

 

[danb35]

What I've read so far says, that it does not need to be redundant because if device fails, the ZIL just goes to the pool where it would be without SLOG anyways.

Correct. But if you're using sync writes heavily, this is very bad.

 

[Exhorder]

Correct. But if you're using sync writes heavily, this is very bad.

Indeed. But at least it (hopefully) won't corrupt my pool.

 

[Exhorder]

Just wanted to add something: I've finally found possibility to add 2 SLOG devices to servers where it actually is not possible. Supermicro offers an AOC-SLG3-2M2 add-on card, that can hold 2 NVMe M.2 SSDs with 110 mm length (e.g. Intel Optane P4801X): https://www.supermicro.com/products/accessories/addon/AOC-SLG3-2M2.cfm

But still I don't know if I really should have 2 SLOG devices ;)

 

[Exhorder]

*bump*