[libvirt] [PATCH 1/2] v3: put dnsmasq parameters into a file instead of the command line

[Gene Czarcinski]

On 10/23/2012 12:01 PM, Laine Stump wrote:
> On 10/23/2012 11:07 AM, Gene Czarcinski wrote:
>> This patch changes the way parameters are passed to dnsmasq.  They are
>> put into a conf-file instead of being on the dnsmasq command line.
> I was thinking about this last night after I learned from you that the
> conf file *isn't* reread when dnsmasq received SIGHUP. That being the
> case, what are the other reasons for switching from commandline to conf
> file? I suppose one is that it eliminates clutter in the ps output (and
> gets it away from prying eyes (since anyone can see the commandline by
> looking in the output of ps -AlF). Another would be it avoids hitting
> the commandline limit (which is rather large, but still exists) when
> there are a lot of srv and txt entries. Anything else?

Not really.  Maybe there is also the fact that I believe this is a 
"cleaner," but not necessarily better, approach.

There is also the possibility that Simon could be convinced to add the 
capability of re-reading configuration files.  I will ask to see what 
his reaction is.  But, his reaction may be: "Sure, go ahead and do that; 
I look forward to your patches."  I am not sure I want to wander in that 
swamp right now.

The most important thing to me is adding the --conf-dir=<directory>.  
This will allow debugging different parameters without having to 
recompile/rebuild the entire libvirt set of packages.  Now, when I first 
did this, I thought that the configuration files in the directory would 
be re-read.  When it did not do that, and, upon doing a little research, 
found that it was working as designed.

Oh, BTW, there is one other little thing about the command line 
parameters.  To support IPv6 completely, the should be 
local=/<reverse-ip6-addr>.ip6.arpa/ parameters and these are very long.  
Putting things into a conf-file just makes it (IMHO) a lot simpler to 
understand.

I do realize that this is open source and I can always "do my own thing" 
and continue using the patches myself.  However, that can mean a lot of 
work just keeping up.

One last observation/question:  Why not?

Gene

>
>> **NOTE ** This has updated the related tests for the new
>> data format, etc.
>>
>> **NOTE** This patch does NOT include specifying interface=
>>
>> The command line now contains --conf-file=<filename> and a new
>> parameter --conf-dir=<directoryname> has been added.
>>
>> The new file and directory are put in the same directory as the
>> leases file.
>> ---
>>   src/network/bridge_driver.c                        | 179 ++++++++++++++-------
>>   src/network/bridge_driver.h                        |   8 +-
>>   tests/networkxml2argvdata/isolated-network.argv    |  24 +--
>>   .../networkxml2argvdata/nat-network-dns-hosts.argv |  14 +-
>>   .../nat-network-dns-srv-record-minimal.argv        |  35 ++--
>>   .../nat-network-dns-srv-record.argv                |  35 ++--
>>   .../nat-network-dns-txt-record.argv                |  29 ++--
>>   tests/networkxml2argvdata/nat-network.argv         |  27 ++--
>>   tests/networkxml2argvdata/netboot-network.argv     |  28 ++--
>>   .../networkxml2argvdata/netboot-proxy-network.argv |  25 +--
>>   tests/networkxml2argvdata/routed-network.argv      |  12 +-
>>   tests/networkxml2argvtest.c                        |  44 +----
>>   12 files changed, 264 insertions(+), 196 deletions(-)
>>
>>
<snip>