[OpenAFS-devel] Patch for XML switch for vos examine

Jeffrey Hutzelman jhutz@cmu.edu
Tue, 20 Apr 2010 15:44:09 -0400


--On Monday, April 19, 2010 07:59:58 PM -0400 Jason Edgecombe 
<jason@rampaginggeek.com> wrote:

> Steve Simmons wrote:
>> On Apr 19, 2010, at 9:54 AM, Jeffrey Altman wrote:
>>
>>
>>> On 4/19/2010 2:17 PM, Sanket Agarwal wrote:
>>>
>>> I would wrap this blob with
>>>
>>> <cell name="cellname">
>>>
>>
>> I'm torn on this one. You can't run vos e on more than one cell at a
>> time, so don't see this as particularly useful in most circumstances.
>> However, if it *is* decided that it should be useful, I'd implement as
>> jaltman suggests - in particular, don't put the cell name inside the
>> volume-specific detail.
>>
>> Similarly, data like server, server IP address and partition is largely
>> redundant when you're dumping many volumes. My xml-foo is amazingly
>> week, but something like
>>
>> <cellname="cellname">
>>    <server=1>
>>        <uuid="xxxx">
>>        <ipv4list>
>>           <ipv4="192.168.1.1">
>>           <ipv4="10.0.0.1">
>>        </ipv4list>
>>        <partitions>
>>            <vicepa="/vicepa">
>>            <vicepb="/vicepb">
>>        </partitions>
>>    </server>
>> </cellname>
>> <volume>
>>    <name>root.cell</name>
>>    <id>536870915</id>
>>    <server=1>
>>    <partition="/vicepa">
>>    ...
>>
>> would eliminate a helluva lot of redundancy. It might also be worthwhile
>> to do this down to a vice partition level, eg,
>>
>> ...
>>        <partitions>
>>            <partition>
>>               <partid="1">
>>               <vicepa="/vicepa">
>>            </partition>
>>            <partition>
>>               <partid=2>
>>               <vicepb="/vicepb">
>>            </partition>
>>        </partitions>
>> ...
>> <volume>
>>    <name>root.cell</name>
>>    <id>536870915</id>
>>    <partid=1>
>>    ...
>>
>>
>>>> why <serv> and not <server>?
>>>>
>>
>> Ditto for partition vs part. Right now he's just using the tags as
>> reported by -format. I suspect that 'serv' vs 'server' and 'part' vs
>> 'partition' was intentional on the original author's part; it does make
>> parsing of the original ascii output easier. I've used that distinction
>> myself for that purpose.
>>
>> Steve_______________________________________________
>> OpenAFS-devel mailing list
>> OpenAFS-devel@openafs.org
>> https://lists.openafs.org/mailman/listinfo/openafs-devel
>>
>>
> would it be better to have <ip version="4" address="192.168.1.1" /> ?
> Would that be better to anticipate ipv6?

No, I don't think so.  IMHO it is better to think of IPv6 as a different 
protocol, rather than as a new version of an existing protocol.