All of lore.kernel.org
 help / color / mirror / Atom feed
From: james.smart@broadcom.com (James Smart)
Subject: [PATCH nvme-cli rfc] fabrics: support default connect/discover args
Date: Wed, 1 May 2019 14:49:48 -0700	[thread overview]
Message-ID: <9614e100-3919-581c-0919-8e263e5633a5@broadcom.com> (raw)
In-Reply-To: <e060385b-7695-d5b1-d4ca-2357e016219d@grimberg.me>



On 4/30/2019 12:34 AM, Sagi Grimberg wrote:
>
>>> Effectively yes, but defargs does not allow 
>>> transport,traddr,trsvcid,nqn
>>> parameters (which don't make sense), and discovery.conf is designed to
>>> hold exactly those. So in practice they should not overlap.
>>>
>> Uh-oh. I _do_ know of cases where discovery.conf is used for 
>> precisely this purpose, ie holding _all_ arguments.
>
> That's fine.. it was never designed for this purpose though...
>
>> And in the absense of any other way existing implementations will be 
>> using a very similar thing.
>
> Again, this would work.
>
>>> Also, we only go to discovery.conf if we did not get a transport+traddr
>>> (which are mandatory). So its not exactly the order you mentioned, more
>>> like:
>>>
>>> defargs.conf
>>> <cmdline>
>>> if transport+traddr not given
>>> discovery.conf
>>>
>> How very curious ...
>> I would have expected that any command-line args would overwrite any 
>> arguments given in the configuration files.
>> But given that discovery.conf can contain several lines I can see the 
>> problem.
>
> Exactly. discovery.conf was simply added such that you can run nvme
> connect-all without explicitly passing parameters so you can build
> auto-connect scripts on top of.

with my suggestion it would be:
 ?? <parse cmdline for connect addressing args>
 ?? if transport+traddr not given:? pull them from discovery.conf note: 
if non-addressing args present, remember them
 ?? parse addressing args vs connectargs.conf? (aka defargs.conf)
 ? <apply non-addressing args from discovery.conf>
 ? <apply cmdline non-addressing args>

-- james

  reply	other threads:[~2019-05-01 21:49 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-29 22:53 [PATCH nvme-cli rfc] fabrics: support default connect/discover args Sagi Grimberg
     [not found] ` <CGME20190429225354epcas4p3d772e0abb1a7ec23c9babd7065844058@epcms2p8>
2019-04-29 23:47   ` Minwoo Im
2019-04-30  0:12 ` Heitke, Kenneth
2019-04-30  5:18   ` Sagi Grimberg
2019-04-30  5:43 ` Hannes Reinecke
2019-04-30  5:54   ` Sagi Grimberg
2019-04-30  6:37     ` Hannes Reinecke
2019-04-30  7:34       ` Sagi Grimberg
2019-05-01 21:49         ` James Smart [this message]
2019-05-01 21:23 ` James Smart
2019-05-01 21:31   ` James Smart
2019-05-01 22:36 ` Arun Easi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9614e100-3919-581c-0919-8e263e5633a5@broadcom.com \
    --to=james.smart@broadcom.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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.