[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha
Subject:    Re: NBD (Network Block Device) internals....
From:       "Stephen C. Tweedie" <sct () redhat ! com>
Date:       2000-03-03 14:45:25
[Download RAW message or body]

Hi,

On Thu, 2 Mar 2000 14:49:31 -0800 (PST), Medi Montaseri
<medi@CyberShell.com> said:

> On Thu, 2 Mar 2000, Mike Tilstra wrote:

>> > In an HA scenario, where the server may move between nodes and we want
>> > the client to reconnect transparently, it will be absolutely essential
>> > to have a high degree of user-space control over the client, including:

> I still don't see why we need a whole process just for that. 

Who is talking about a whole process?   It doesn't matter what the API
is, as long as the HA manager can give the appropriate commands to the
driver when membership changes happen.

> Assuming existance of some cluster manager (monitor), a file should be
> used for the NBD client to read its instruction regarding IP, Port,
> Size of the exporter. 

The trouble is, you need to be able to change this information
dynamically without removing the driver.

> A different design would be to to have the nbd-client read from a file
> due to recipt of a signal or read from a named pipe. 

An ioctl() would be much more appropriate for a kernel-mode nbd client. 

--Stephen

------------------------------------------------------------------------------
Linux HA Web Site:
  http://linux-ha.org/
Linux HA HOWTO:
  http://metalab.unc.edu/pub/Linux/ALPHA/linux-ha/High-Availability-HOWTO.html
------------------------------------------------------------------------------

[prev in list] [next in list] [prev in thread] [next in thread]