Opened 9 months ago

Last modified 9 months ago

#1748 new defect

Inconsistency in nvme output & recommendation

Reported by: Mark Thompson Owned by:
Priority: minor Milestone: undecided
Component: smartctl Version: 7.4
Keywords: nvme Cc:

Description

Smartctl 7.4 output for NVMe SMART Self-test Logs is not consistent between HDD or SSD output format, specifically the Self-test Log "Num" column is not formatted to start with "#" as the HDD and SSD outputs are formatted. Additionally the numbering starts at "0" on an nvme output where everything else starts at "1". As a side note: I also echo the problem in ticket 1741.

Attachments (1)

nvme_normal_outut.txt (3.1 KB ) - added by Mark Thompson 9 months ago.

Download all attachments as: .zip

Change History (4)

by Mark Thompson, 9 months ago

Attachment: nvme_normal_outut.txt added

comment:1 by Mark Thompson, 9 months ago

Summary: Inconsistency in outputInconsistency in nvme output & recommendation

I do not program in C++ but I hope to have a tiny grasp on it. If my recommendation is bad, sorry.

In version 7.5 src for nvmeprint.ccp, line 643:

jout("%2u %-17s %-33s %9" PRIu64 " %12s %5s %3s %3s %4s\n", i, t, s, poh, lb, ns, sg, st, sc);

Change to (add hash at beginning):

jout("#%2u %-17s %-33s %9" PRIu64 " %12s %5s %3s %3s %4s\n", i, t, s, poh, lb, ns, sg, st, sc);

comment:2 by Christian Franke, 9 months ago

Keywords: nvme added
Milestone: undecided

In general, ATA/SATA, SCSI/SAS and NVMe diagnostic information differ considerably by design and so do smartctl outputs.

The start with "0" is intentional because it is the index of the table entry. Aligning with the 20+ years old ATA/SATA format won't ease table parsing much because the remaining columns also differ.

Leaving ticket open as undecided for now.

comment:3 by Mark Thompson, 9 months ago

I can understand that point about the number itself, '0' is perfectly fine, however what about the '#' preceding the test number? I saw an consistency issue that was introduced in version 7.4 just for nvme and thought I'd recommend the change to maintain consistency amongst the Selt-test Logs.

Note: See TracTickets for help on using tickets.