Hrvoje Crvelin wrote:
>
> > Hot line said me that it could be a problem.
> This may cause some trouble. I do not agree with
> arguments You gave. They do stand for 6.0.x and
> 6.1.x (and there is no problem up to my knowledge),
> but difference between 6.x and 5.x is rather big
> so mixing those may result in problems. Legato is
> not some kind of isolated example here - almost all
> software companies to this. And there are good
> reasons for that if You think for a moment.
I don't want to start a flame war. I only want to tell why I'm not happy
with networker.
OK ! I think only that if there are incompatibilities, software should
detect it and show an alert message. This is how quality software
companies do !
Let's talk about other examples :
- recently, I needed to restore two disk partitions (Solaris) - a 9 GB
partitions, 50 % full - home dir of 260 users on each partition.
So I launched nwrecover and select (mark) the top of the partition. When
I did this, nwrecover froze. I waited some 15 minutes and killed it and
restarted again for some half hour. As I saw the same behaviour, I
changed the way to restore.
So intead of selecting the top of the partition, I went into the
partition and tried to select all directories inside the partition.
So when I selected each directory, It tooks some time to show the mark
telling the selection was done. I did this for each 260 directories in
each partition.
What's wrong ?
- There is no selection tool to select a group of directories. I needed
to repeat the same operation 520 times.
- There is no message indicating the operation progression. If an
operation takes some time, software usually shall set up some way to
indicating task progression.
I finaly understood that when I selected the top of the partition,
nwrecover was realy working, and not frozen.
What's wrong ? It could be a good thing to to setup some way to tell
user that the process isn't dead, but working. Something such a bar
indicating % of work done, or at least a message each minute saying
"Hello, I'm alive. Don't kill me, please !".
I have many examples like this one. It doesn't seems to me necessary to
tell you all of them.
>
> Note that keeping 5.5.x server at the time when
> 5.5.x is EOL with fresh clients is bad admin job
Thanks for your opinion. 8-)
> as well. While NetWorker is client-server SW,
> most of the operations do happen on server side.
> This server side has been enhanced great deal
> (RFEs and bug fixes) so I would really urge You
> to upgrade. Perhaps then You would be less tired
> of NetWorker problems You mentioned.
>
> Back to original question. You may go to 6.1.x
> w/o any problem. To be 100% sure, make following:
> 1) Configure dummy machine
> 2) Stop original backup server and copy mm, index and res
> to the dummy machine
> 3) Start original one
> ---
> At this point original should continue to work
> and dummy one should be not be on net. Name the
> dummy machine as original backup machine and install
> same version of NetWorker as You have on original.
> Make an update test then which in short should be:
> 1) eject all tapes from drives
> 2) disable drives
> 3) disable groups
> 4) run nsrck -F for all clients
> 5) clear $NSR/tmp
> 6) uninstall (partial) all software
> 7) install new one
> - I would recommend to use tail -f daemon.log, top and
> ps in loop to check the progress here; on vanila Windows
> You have only task manager there (but You may use some
> other utils laying around)
> 8) check the logs
> 9) enable disabled (skip that when doing test)
> 10) make a test backup of dummy machine (You may use
> file device)
>
> The goal of this test is to see:
> a) if everything works fine
> b) measure the time needed for operation (aprox. value since
> You would need to replicate SW and HW environment for real
> figures) which should give You an idea of amount of time
> needed for operation.
>
> I'm quite sure I forgot something, but trying this on Your
> own should give You an idea.
>
> Cheers,
> Hrvoje Crvelin
>
> --
> Note: To sign off this list, send a "signoff networker" command via email
> to listserv AT listmail.temple DOT edu or visit the list's Web site at
> http://listmail.temple.edu/archives/networker.html where you can
> also view and post messages to the list.
> =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
--
---------------------------------------------------------------
Jose Marcio MARTINS DA CRUZ Tel. :(33) 01.40.51.93.41
Ecole Nationale Superieure des Mines de Paris
Centre de Calcul http://j-chkmail.ensmp.fr
60, bd Saint Michel http://www.ensmp.fr/~martins
75272 - PARIS CEDEX 06 mailto:Jose-Marcio.Martins AT ensmp DOT fr
---------------------------------------------------------------
--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
|
