adding system snapshots to ubuntu standard install

Fri Aug 18 21:48:14 UTC 2006

hello friends,

im trying to add a kind of system snapshot feature in my standard
desktop install of ubuntu dapper.

basically what i intend to do is that i freeze my system state at some
point in time. after this whatever i do to the system is recorded as a
delta somewhere else.

this is something like vmware snapshot feature but off the real
hardware without using vmware.

its actually working and now when im typing this post, all the changes
are being recorded in the delta. i can install new packages, change
configuration blah blah and it will go in the delta. but im observing
some errors which irritates me

so first of all, how im doing it. im using unionfs to enable this
snapshot feature. here is what i have done...

1. i have ubuntu dapper installed on /dev/hda1 (ext3)

2. i have another ext3 partition on /dev/hda2. this is the one im using
to store all the delta, so my original install in /dev/hda1 wont be
touched at all

3. i modified the default initrd to mount this extra partition
(/dev/hda2) on /extra_root

4. the default scripts in init have already mounted my actual root
(/dev/hda1) on /root

5. i then create a unionfs on /virtual_root by combining /extra_root
and /root, making /root as a readonly branch and /extra_root as
writable branch

6. after this i just use /virtual_root to call the real init _and the
system boots on top of my union, and hence recording all the changes in
/dev/hda2

my unionfs is configured like this

/dev/hda1 on /root (in initrd)

/dev/hda2 on /extra_root (in initrd, done by me)

command line for mounting unionfs is

mount -t unionfs -o dirs=/extra_root=rw:/root:ro unionfs /virtual_root

so one of the error that im getting is just after the real init is
called, here it is... i have googled a bit and its in unionfs but the
thing is that its working even with message but then why and whats the
reason for this message... i was earlier getting this error when the
branches of my unionfs were overlapping but it is not the case
anymore...

[4294705.764000] Unable to handle kernel NULL pointer dereference at
virtual address 00000050

[4294705.764000]  printing eip:

[4294705.764000] c01651d4

[4294705.764000] *pde = 00000000

[4294705.764000] Oops: 0000 [#1]

[4294705.764000] PREEMPT 

[4294705.764000] Modules linked in: ext3 jbd ide_generic ehci_hcd
uhci_hcd usbcore ide_cd cdrom ide_disk piix generic thermal processor
fan unionfs fbcon tileblit font bitblit softcursor capability
commoncap

[4294705.764000] CPU:    0

[4294705.764000] EIP:    0060:[<c01651d4>]    Not tainted VLI

[4294705.764000] EFLAGS: 00010246   (2.6.15-23-386) 

[4294705.764000] EIP is at bio_get_nr_vecs+0x4/0x60

[4294705.764000] eax: 00000000   ebx: 00000000   ecx: 00000001   edx:
00000000

[4294705.764000] esi: c13ea9a0   edi: dfca2c3c   ebp: 00000001   esp:
dfb77d84

[4294705.764000] ds: 007b   es: 007b   ss: 0068

[4294705.764000] Process readahead-list (pid: 2102, threadinfo=dfb76000
task=dfabd550)

[4294705.764000] Stack: c0185b0b 00000000 dfcb2900 ded28424 dfb77dc8
0000000c dfca2cd4 00000001 

[4294705.764000]        00000001 00000000 00000001 00000000 00000001
00000000 00000001 dfca2c3c 

[4294705.764000]        00000000 dfb77e3c 0000000d dffac940 dfb76000
def8005c 00000046 00000000 

[4294705.764000] Call Trace:

[4294705.764000]  [<c0185b0b>] do_mpage_readpage+0x40b/0x4c0

[4294705.764000]  [<c0149001>] kmem_cache_alloc+0x31/0x40

[4294705.764000]  [<c01dafd5>] radix_tree_node_alloc+0x15/0x60

[4294705.764000]  [<c01db29f>] radix_tree_insert+0xff/0x140

[4294705.764000]  [<c013fa26>] add_to_page_cache+0x46/0xb0

[4294705.764000]  [<c0185c5a>] mpage_readpages+0x9a/0x130

[4294705.764000]  [<e090f610>] ext3_get_block+0x0/0xc0 [ext3]

[4294705.764000]  [<e085fbb2>] unionfs_permission+0x72/0xd0 [unionfs]

[4294705.764000]  [<c0143f78>] prep_new_page+0x68/0x80

[4294705.764000]  [<c0144537>] buffered_rmqueue+0xc7/0x250

[4294705.764000]  [<c0147117>] read_pages+0x27/0x130

[4294705.764000]  [<e090f610>] ext3_get_block+0x0/0xc0 [ext3]

[4294705.764000]  [<c014489e>] __alloc_pages+0x4e/0x2e0

[4294705.764000]  [<c0147391>] __do_page_cache_readahead+0x171/0x1d0

[4294705.764000]  [<c0147433>] force_page_cache_readahead+0x43/0xa0

[4294705.764000]  [<c0140ab9>] do_readahead+0x39/0x40

[4294705.764000]  [<c0140b36>] sys_readahead+0x76/0x90

[4294705.764000]  [<c010302b>] sysenter_past_esp+0x54/0x79

[4294705.764000] Code: 5a 5b 5e 5f 5d c3 53 8b 44 24 04 50 e8 f6 69 06
00 5d 58 8b 53 10 eb e1 53 8b 7c 24 04 57 e8 e4 69 06 00 59 5e eb d7 8b
44 24 04 <8b> 40 50 8b 48 34 8b 81 3c 01 00 00 25 ff ff 00 00 c1 e0 09
8d 

the good thing is that its working even with this error, but i dont
like errors..

the other problem is that how should i unmount all the filesystems at
shutdown.

thanks in advance.

regards,

- kamal n

-- 
nijhawank