[linux-lvm] metadata format processing

[linux-lvm] metadata format processing

[Dieter Stüken]

I want to develop a tool to get a graphical representation of my LVM configuration.
I thought about parsing the output of lvm commands, or even using the C-API directly.
Unfortunately I did not found a way to enforce read only access when using the API.
All user space tools have to run as root and may modify/destroy my LVM. So I think
they are a bad choice to build any CGI-scripts upon. Finally I discovered the external
metadata format, which describes any aspects of my LVM in detail. After I even found
a parser for this syntax within the LVM libs, I was able to convert the metadata
into XML. Once having XML it should be possible to transfer any informations towards
HTML or SVG.

My very first try looks like this:


<?xml version="1.0" encoding="ISO-8859-1"?>
<lvm>
   <contents>Text Format Volume Group</contents>
   <version>1</version>
   <description>Created *after* executing &apos;lvextend /dev/raster/eglv -l 7452 /dev/sdg1&apos;</description>
   <creation_host>linux2</creation_host>
   <creation_time>1124130973</creation_time>
   <group name="raster">
     <id>6FmvZu-mLHD-AIQV-jnFu-hMUN-oTwe-X2zCRm</id>
     <seqno>54</seqno>
     <status>RESIZEABLE</status>
     <status>READ</status>
     <status>WRITE</status>
     <extent_size>65536</extent_size>
     <max_lv>255</max_lv>
     <max_pv>255</max_pv>
     <group name="physical_volumes">
       <group name="pv0">
         <id>ruo3gO-jHR7-QV3Z-rUa4-jYnF-GyAn-uUR3vA</id>
         <device>/dev/sde1</device>
         <status>ALLOCATABLE</status>
         <pe_start>384</pe_start>
         <pe_count>7452</pe_count>
       </group>
       <group name="pv1">
         <id>ARjpMz-A6iP-8x3t-7aCg-CCSH-kZpo-vt1KPD</id>
         <device>/dev/sdd1</device>
         <status>ALLOCATABLE</status>
         <pe_start>384</pe_start>
         <pe_count>7452</pe_count>
       </group>
     </group>
     <group name="logical_volumes">
       <group name="eglv">
         <id>ylp7ol-r6wT-e2Py-4KZ9-cgjU-6fU0-luciPx</id>
         <status>READ</status>
         <status>WRITE</status>
         <status>VISIBLE</status>
         <segment_count>1</segment_count>
         <group name="segment1">
           <start_extent>0</start_extent>
           <extent_count>7452</extent_count>
           <type>striped</type>
           <stripe_count>1</stripe_count>
           <stripes>pv0</stripes>
           <stripes>0</stripes>
         </group>
       </group>
     </group>
   </group>
</lvm>

I manually shortened it, just as an example...

I used the parser to get it into a "struct config_tree" and developed
some procedure to dump it out as XML. There are some aspects that can be
discussed (why are all nesting elements simply called <group> whereas
arrays are not grouped ...)

I'm just about to learn XSLT to get this processed further.

Is some one interested to participate?

Dieter.
-- 
Dieter St�ken, con terra GmbH, M�nster
     stueken@conterra.de
     http://www.conterra.de/
     (0)251-7474-501