Backup Central

I'm restoring some files using amrestore on the client machine, but due
to the size of the tape being restored from (100GB) I usually hit a
timeout on the client side while waiting for amrestore on the server to
find the correct image.

So, short of recompiling to increase READ_TIMEOUT like amrecover
suggests, is there any simple way to get the restore process to work
faster? Is there a neat way of keeping a precomputed index of the tape
somewhere so that amrestore can seek to the right image straight away?

Or is the only solution to manually position the tape just before the
right image before typing 'extract' at the client?

I think last time I attempted this I gave up trying to use amrecover and
used amrestore to extract the full image, and then copied it to the
client where I just used tar to recover the files I needed, instead of
using the amanda client programs.

--
jaq < at > spacepants.org http://spacepants.org/jaq.gpg

This one time, at band camp, Jamie Wilkinson wrote:
I'm restoring some files using amrestore on the client machine, but due

I mean, amrecover on the client machine.

--
jaq < at > spacepants.org http://spacepants.org/jaq.gpg

Jamie Wilkinson wrote:

So, short of recompiling to increase READ_TIMEOUT like amrecover
suggests, is there any simple way to get the restore process to work
faster? Is there a neat way of keeping a precomputed index of the tape
somewhere so that amrestore can seek to the right image straight away?

In amanda 2.4.3 these two parameters in amanda.conf were introduced:

amrecover_do_fsf yes
amrecover_check_label yes

You need both set to yes if using a vanilla 2.4.4p2 or earlier (a little
bug which was fixed a few weeks later, and is in all current snapshots).


--
Paul Bijnens, Xplanation Tel +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUM Fax +32 16 397.512
http://www.xplanation.com/ email: Paul.Bijnens < at > xplanation.com
***********************************************************************
* I think I've got the hang of it now: exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit, ZZ, :q, :q!, M-Z, ^X^C, logoff, logout, close, bye, /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt, abort, hangup, *
* PF4, F20, ^X^X, :, KJOB, F14-f-e, F8-e, kill -1 $$, shutdown, *
* kill -9 1, Alt-F4, Ctrl-Alt-Del, AltGr-NumLock, Stop-A, ... *
* ... "Are you sure?" ... YES ... Phew ... I'm out *
***********************************************************************

This one time, at band camp, Paul Bijnens wrote:
Jamie Wilkinson wrote:

So, short of recompiling to increase READ_TIMEOUT like amrecover
suggests, is there any simple way to get the restore process to work
faster? Is there a neat way of keeping a precomputed index of the tape
somewhere so that amrestore can seek to the right image straight away?

In amanda 2.4.3 these two parameters in amanda.conf were introduced:

amrecover_do_fsf yes
amrecover_check_label yes

You need both set to yes if using a vanilla 2.4.4p2 or earlier (a little
bug which was fixed a few weeks later, and is in all current snapshots).

Thanks for this, just what I was looking for.

--
jaq < at > spacepants.org http://spacepants.org/jaq.gpg

This one time, at band camp, Paul Bijnens wrote:
Jamie Wilkinson wrote:

So, short of recompiling to increase READ_TIMEOUT like amrecover
suggests, is there any simple way to get the restore process to work
faster? Is there a neat way of keeping a precomputed index of the tape
somewhere so that amrestore can seek to the right image straight away?

In amanda 2.4.3 these two parameters in amanda.conf were introduced:

amrecover_do_fsf yes
amrecover_check_label yes

You need both set to yes if using a vanilla 2.4.4p2 or earlier (a little
bug which was fixed a few weeks later, and is in all current snapshots).

Does the client need to be >= 2.4.3 also?

I have a client running Red Hat 8.0 (amanda-client 2.4.2p2-9) and a
server with RHEL-3ES (amanda-server 2.4.4p1-0.3E) and those two settings
above didn't stop the timeout.

--
jaq < at > spacepants.org http://spacepants.org/jaq.gpg

Jamie Wilkinson wrote:

This one time, at band camp, Paul Bijnens wrote:

Jamie Wilkinson wrote:

So, short of recompiling to increase READ_TIMEOUT like amrecover
suggests, is there any simple way to get the restore process to work
faster? Is there a neat way of keeping a precomputed index of the tape
somewhere so that amrestore can seek to the right image straight away?

In amanda 2.4.3 these two parameters in amanda.conf were introduced:

amrecover_do_fsf yes
amrecover_check_label yes

You need both set to yes if using a vanilla 2.4.4p2 or earlier (a little
bug which was fixed a few weeks later, and is in all current snapshots).

Does the client need to be >= 2.4.3 also?

I have a client running Red Hat 8.0 (amanda-client 2.4.2p2-9) and a
server with RHEL-3ES (amanda-server 2.4.4p1-0.3E) and those two settings
above didn't stop the timeout.

No the client is not relevant, I believe. (not yet verified in the sources)

Is the server doing an "mt fsf" or is it just reading the tape up to the
matching backup image?

After how long does it time out? Is there a firewall in between maybe?


--
Paul Bijnens, Xplanation Tel +32 16 397.511
Technologielaan 21 bus 2, B-3001 Leuven, BELGIUM Fax +32 16 397.512
http://www.xplanation.com/ email: Paul.Bijnens < at > xplanation.com
***********************************************************************
* I think I've got the hang of it now: exit, ^D, ^C, ^\, ^Z, ^Q, F6, *
* quit, ZZ, :q, :q!, M-Z, ^X^C, logoff, logout, close, bye, /bye, *
* stop, end, F3, ~., ^]c, +++ ATH, disconnect, halt, abort, hangup, *
* PF4, F20, ^X^X, :, KJOB, F14-f-e, F8-e, kill -1 $$, shutdown, *
* kill -9 1, Alt-F4, Ctrl-Alt-Del, AltGr-NumLock, Stop-A, ... *
* ... "Are you sure?" ... YES ... Phew ... I'm out *
***********************************************************************

This one time, at band camp, Paul Bijnens wrote:
Is the server doing an "mt fsf" or is it just reading the tape up to the
matching backup image?

It's not spawning a "mt fsf" when I do a ps, as far as I can tell.

If I strace the amrestore process I see it doing the following ioctls
between file marks:

ioctl(3, MTIOCTOP or SNDCTL_MIDI_MPUMODE or MGSL_IOCPARAMS

"history" in amrecover lists the file marks that each tape has that
image at, so I assume I'm not missing any indexes that amrestore will
need.

After how long does it time out? Is there a firewall in between maybe?

No firewall on the backup network, at least I've been able to back up
each night, and do manual restores before.

The timeouts usually occur after a half hour, I haven't timed it though.

--
jaq < at > spacepants.org http://spacepants.org/jaq.gpg