From: "Bryn M. Reeves" <bmr@redhat.com>
To: device-mapper development <dm-devel@redhat.com>
Cc: John Strange <johnstra10@gmail.com>
Subject: Re: noob usage question
Date: Mon, 12 Aug 2013 12:40:07 +0100 [thread overview]
Message-ID: <5208C997.503@redhat.com> (raw)
In-Reply-To: <CAH1aXXZHc8rYQUfbQsAXgrjrryLvZb-NfyTEdTx+a9oJSHwc-g@mail.gmail.com>
On 08/10/2013 03:19 AM, John Strange wrote:
> Yes, I think I do want to create partitions. My expectation is to create a
> mapping based on one or more raw physical device (luns). There are a number
> features I would like implement behind my mapping target driver. My
> understanding is that I can treat the mapping (the target device) as if it were
> a raw disk and that I can format it as appropriate for whatever workload will
> use it.
For typical uses you would format the raw physical device and then use
the kpartx device-mapper application to create linear maps for each
partition contained on the device.
For e.g. this is how users work with partitions on multipath devices
managed by multipath-tools and device-mapper.
> For instance... I need to present a device (it would appear to the OS as a raw
> block device) composed of ranges of blocks from different devices - spinning
> media, ssd, PCIe flash, etc. My target driver can guarantee a QOS (throughput,
> latency, combo) and it will service read requests across the range of physical
> media comprising the target.
>
> This target device would appear to the OS as a block device. The workload that
> uses the device might require a filesystem on it, so I need to mount a FS.
Still not sure why you would want to place partitions on the linear maps
but you can do it if you really want to:
http://paste.fedoraproject.org/31461/76307438/
First example is creating and partitioning a linar mapping (I used a
loop device as the backing but it will work with any block device).
Following that is creating a file system on a partitioned linear device,
and then on the whole linear device.
There's no problem with either operation.
> That's what I'm aiming for. I ran into problems trying to format my target
> device so I created a linear mapping and it behaved the same way... could not
> put a partition on it or format it.
You do not need a partition to be able to format or use a device-mapper
device - dm's devices do not support partitions (which is why you have
to use the userspace kpartx - it reads partition tables and creates
equivalend DM linear mappings for them).
It sounds like something else is going on in your case - what errors do
you get when you try to use the linear devices (busy for e.g.)?
Regards,
Bryn.
prev parent reply other threads:[~2013-08-12 11:40 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-08-09 21:01 noob usage question John Strange
2013-08-09 22:22 ` Brassow Jonathan
2013-08-10 2:19 ` John Strange
2013-08-12 11:40 ` Bryn M. Reeves [this message]
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=5208C997.503@redhat.com \
--to=bmr@redhat.com \
--cc=dm-devel@redhat.com \
--cc=johnstra10@gmail.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.