All of lore.kernel.org
 help / color / mirror / Atom feed
* Re: [SPDK] analyzing spdk error messages
@ 2016-10-12 16:40 Walker, Benjamin
  0 siblings, 0 replies; 3+ messages in thread
From: Walker, Benjamin @ 2016-10-12 16:40 UTC (permalink / raw)
  To: spdk

[-- Attachment #1: Type: text/plain, Size: 1703 bytes --]

On Fri, 2016-10-07 at 17:37 +0000, txcy uio wrote:
> Hello
> 
> 
> What is the correct way of analyzing spdk error message (as below which was
> encountered against a spdk writev cmd) to quickly know what went wrong?

These messages are dumping compact notes about the command that was sent and the
completion returned. The values are pulled directly from the specification - see
section 4.6 of NVMe 1.2.1 (http://www.nvmexpress.org/wp-content/uploads/NVM_Expr
ess_1_2_1_Gold_20160603.pdf) for a full description.
> 
> 
> WRITE sqid:2 cid:125 nsid:1 lba:4194304 len:16

WRITE - Command Type
sqid - Submission Queue Id (2)
cid - Command Id (125). This
is an arbitrary number we picked for this command. We match completions to
commands by this number.
nsid - Namespace Id (1)
lba - Logical Block Address
(4194304)
len - Number of logical blocks (16)

> WRITE FAULTS (02/80) sqid:2 cid:125 cdw0:0 sqhd:002a p:1 m:1 dnr:0

WRITE FAULTS - The error encountered. The (02/80) is (Status Code Type, Status
Code Value). 02 means media error type, and 80 in the media error category means
write fault. These values are in hex.
sqid - Submission Queue Id (2)
cid - Command
Id (125). Note how it matches up to the command sent.
cdw0 - Command DWORD 0 (0).
Dumped as hex.
sqhd - Submission Queue Head (2a). The current location of the
device's submission queue head index.
p - Phase bit (1)
m - More (1). If set to 1,
there is additional error information available that can be obtained by sending
a Get Log Page command.
dnr - Do Not Retry (0).

This error seems to indicate that there was a write failure in the media on your device. 

> 
> 
> 
> 
> 
> --Tyc
> 

^ permalink raw reply	[flat|nested] 3+ messages in thread
* Re: [SPDK] analyzing spdk error messages
@ 2016-10-12 16:43 Daniel Verkamp
  0 siblings, 0 replies; 3+ messages in thread
From: Daniel Verkamp @ 2016-10-12 16:43 UTC (permalink / raw)
  To: spdk

[-- Attachment #1: Type: text/plain, Size: 1991 bytes --]

On 10/12/2016 09:40 AM, Walker, Benjamin wrote:
> On Fri, 2016-10-07 at 17:37 +0000, txcy uio wrote:
>> Hello
>>
>>
>> What is the correct way of analyzing spdk error message (as below which was
>> encountered against a spdk writev cmd) to quickly know what went wrong?
>
> These messages are dumping compact notes about the command that was sent and the
> completion returned. The values are pulled directly from the specification - see
> section 4.6 of NVMe 1.2.1 (http://www.nvmexpress.org/wp-content/uploads/NVM_Expr
> ess_1_2_1_Gold_20160603.pdf) for a full description.
>>
>>
>> WRITE sqid:2 cid:125 nsid:1 lba:4194304 len:16
>
> WRITE - Command Type
> sqid - Submission Queue Id (2)
> cid - Command Id (125). This
> is an arbitrary number we picked for this command. We match completions to
> commands by this number.
> nsid - Namespace Id (1)
> lba - Logical Block Address
> (4194304)
> len - Number of logical blocks (16)
>
>> WRITE FAULTS (02/80) sqid:2 cid:125 cdw0:0 sqhd:002a p:1 m:1 dnr:0
>
> WRITE FAULTS - The error encountered. The (02/80) is (Status Code Type, Status
> Code Value). 02 means media error type, and 80 in the media error category means
> write fault. These values are in hex.
> sqid - Submission Queue Id (2)
> cid - Command
> Id (125). Note how it matches up to the command sent.
> cdw0 - Command DWORD 0 (0).
> Dumped as hex.
> sqhd - Submission Queue Head (2a). The current location of the
> device's submission queue head index.
> p - Phase bit (1)
> m - More (1). If set to 1,
> there is additional error information available that can be obtained by sending
> a Get Log Page command.
> dnr - Do Not Retry (0).
>
> This error seems to indicate that there was a write failure in the media on your device.

Write Fault could also indicate that you passed an incorrect memory 
address and the DMA transfer failed.

Note that the NVMe writev callback requires a physical address, not a 
virtual address.


^ permalink raw reply	[flat|nested] 3+ messages in thread
* [SPDK] analyzing spdk error messages
@ 2016-10-07 17:37 txcy uio
  0 siblings, 0 replies; 3+ messages in thread
From: txcy uio @ 2016-10-07 17:37 UTC (permalink / raw)
  To: spdk

[-- Attachment #1: Type: text/plain, Size: 288 bytes --]

Hello

What is the correct way of analyzing spdk error message (as below which was
encountered against a spdk writev cmd) to quickly know what went wrong?

WRITE sqid:2 cid:125 nsid:1 lba:4194304 len:16
WRITE FAULTS (02/80) sqid:2 cid:125 cdw0:0 sqhd:002a p:1 m:1 dnr:0


--Tyc

[-- Attachment #2: attachment.html --]
[-- Type: text/html, Size: 406 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2016-10-12 16:43 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-10-12 16:40 [SPDK] analyzing spdk error messages Walker, Benjamin
  -- strict thread matches above, loose matches on Subject: below --
2016-10-12 16:43 Daniel Verkamp
2016-10-07 17:37 txcy uio

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.