Best way to mirror HP-UX from one server to another

[Bram Vermeulen]

Greetings,

I'm burdened with the task to make sure we have a good solution for our HP-UX server concerning the BCP. The goal is to have a HP-UX server running on a different location, on a different IP range but in the same domain.

I've considered using Data Protector to backup the data from our live HP-UX server twice a day and restoring it to the other server, but the problem is we have two different cell managers running, one in the other location and one at the live servers. I'm sure it could work if I make one of the cell managers manager-of-managers but I'd rather not do that if there's a better solution.

Another idea is using ignite-ux to make recovery files. But for this I need to make the other server a part of the "ignite network" with my live HP-UX server as a master, right?

Is there something I am not considering that would be an easier solution? I know HP StorageMirroring doesn't work for HP-UX. Is MirrorDisk/UX something to consider? Or is Ignite-UX the best way to go in my situation?

[Geoff Wild · Accepted solution]

Well...first off...no one really DR's a "server" - you DR your data/app/db....

So I wouldn't use ignite for that....

Mirroring has limited distance...

A more expensive approach is to use utilities like EMC SRDF or HP's Business Copy - to do SAN Frame to Frame copying...

A lot of this design depends on the type of data...

Example - if it is just a samba server - why not use rdist/rsync?

If a db - you could always copy the db out to remote sight - then just continuously roll it forward with archive logs...

I don't know Data Protector - but if it is similar to Netbackup, then you couls always "import" the archive from one server to another...then restore the data to the remote server...

Rgds...Geoff

[OldSchool]

Well...i'm reasonably sure that if you go the DP route, you would have to import the media on the destination side, then restore, which will extend the amount of time taken. Besides, this is not a task that DP was originally intended for

Ignite is intended as an OS restore only...again not the proper product for the job.

So...have you looked at either "rdist" (supplied w/ HP-UX) or "rsync" (gnu/fsf?). These can sync files / directories between the boxes.

I believe rsync is smart enough to only send the changed "parts" of a file, rather then the entire thing.

[Bram Vermeulen]

Sorry about the unfortunate naming of the thread, ofcourse I meant to mirror the data and applications. :)

So yes, it's a samba server. My knowledge about this is pretty limited because I work in a pretty small enterprise (~50 employees) so my job description is pretty wide, resulting in limited knowledge about more detailed stuff. After looking up your suggestions rdist/rsync is surely the way I want to go with this situation. The reasons why I mentioned DP and ignite was because I've crossed paths with these programs in the past when I was looking at backupping and DR of a server (if the server isn't completely destroyed ;)).
Since we only have around 50 users the more expensive approaches do sound very interesting but I doubt it's worth the costs given our needs.

Thanks for the fast answers, I'll do the neccesary research into rdist and rsync now :)

[Rita C Workman]

Hi Bram,

If the goal is for "you" to have two sites than, has Geoff has mentioned, you replicate your disk (i.e. arrays) from site-2-site. EMC or HP both offer good products.
To failover from server to server, you can sound like a good fit for a Continental Cluster. For MC/SG it's more than just pkg failover - it's being able to go as far as site failover.

Now, you could do something like your own in-house subscription DR Plan. If your primary site is totally down....You could use your production ignite tapes to rebuild at the Second Site. And you'd have to make adjustments for that site specific differences during the ignite (DNS,IP...) AND you would have to have something to either rebuild your primary cell manager Data Protector somewhere at the second site, or as was mentioned import your data via tape to the existing cell manager at the second site. Ouch....cause that will take alot of time to import THEN restore.

For DR Planning.....first know your budget allowance, then you know how "creative" you have to be. And also important is what amount of time to be back up is expected by the business side.

Hope my ramblings help,
Rgrds,
Rita

[Geoff Wild]

Here's how I use rdist to sync the home dirs of my SAP admin account:

Root's cron:

05 01 * * * /app/admin/drp/update-binaries.sh


# cat /app/admin/drp/update-binaries.sh
#! /bin/sh

# Keep the DRP copy of the vgpadm up-to-date.
# Currently the files are in:
#
# /home/ipradm/DR
#
# See the rdist(1M) distfile for a list of exclusions.

DRPDIR=/app/admin/drp
DRPHOST=svr031

HOST=`/usr/bin/uname -n`
IPR=`cmviewcl |grep iprdbci |awk '{print $5}'`
if [ "$HOST" = "$IPR" ]
then
mount | grep /home > /dev/null 2>&1
if [ $? -eq 0 ]
then
for vg in `ls -d /dev/vg6* |awk -F/ '{print $3}'`
do
/usr/sbin/vgexport -p -s -v -m /home/ipradm/DR/$vg.map /dev/$vg
done
cp -p /etc/cmcluster/IPRDBCI/iprdbci.cntl /home/ipradm/DR/
cp -p /etc/cmcluster/IPRDBCI/iprdbci.cntl.full /home/ipradm/DR/
cp -p /etc/cmcluster/IPRDBCI/iprdbci.cntl.mountonly /home/ipradm/DR/
( su - ipradm -c "rdist -f $DRPDIR/distfile ipradm"; ) 2>&1 |\
tee $DRPDIR/drp.log 2>&1 |\
mailx -s "ipradm DRP rdist output" sysadm
fi
fi



# cat distfile
IPRDR = ( svr031 )

#
# File systems to be copied over to the DR host.
# Don't use -R in install - so as not to remove files on destination host
IPRADM = /home/ipradm

ipradm: ( ${IPRADM} ) -> ( ${IPRDR} )
install -w ;




Rgds...Geoff

[OldSchool]

"Sorry about the unfortunate naming of the thread, ofcourse I meant to mirror the data and applications. :)"

not to worry. appears that rsync would be a good fit for this.

best of luck