Network Management

I am looking to upgrade our Airwave system to 64bit.  We are currently running v7.3.8 32bit.  I checked all the prerequisites in terms of CPU and what not.  We are a candidate for the upgrade.  I am reading a KB article for instructions (http://www.airwave.com/support/knowledge-base/?sid=50140000000aOby ) .  My question is, do I have to do the upgrade to 64bit using the EXACT same version  we are on (7.3.8 but 64bit)?

Yes, you want to maintain the AMP version during the OS upgrade.  You will want to take your backup off server, reinstall with 7.3.8 64-bit, and then restore the backup.  Once restored, verify that the data has restored properly - check a few you device count, view a few of your devices, check VisualRF, check RAPIDS.  After OS upgrade, you can continue to upgrade AMP as normal (you can upgrade directly to 7.5.6 from 7.3.8).

whats the command to restore the backup?

Here's how to restore a backup if the backup is named nightly_data.tar.gz and it is location in folder /tmp.

# screen

Force of habit for any operation that I want to run without monitoring the CLI

# amp_disable

Another force of habit, but it makes sure that nothings still writing to the database as you're trying to restore

# amp_restore -d /tmp/nightly_data.tar.gz

As restore is going on, I tend to detach myself from the session (in screen, ctrl+a then d).  And then when I want to get back into that screen, if it's the only screen -> 'screen -r', if there are several screens, then you may need to do 'screen -ls' and then 'screen -r ####' where the hash is the pid number.  Once restore completes, it will still be in amp_disable state, so 'amp_enable' and then wait 5-10 minutes for the GUI to become available.

ok, got everything back to normal.  Took a backup and offloaded locally.  Do I just copy the 64bit file to the root and perform start_amp_upgrade from CLI?  the KB article says fresh install?  "Please download the 7.3.8 64 bit ISO image from the link given below and perform a fresh installation:"

Major OS upgrades cannot be done using a simple tarball package.  The OS upgrade path requires you to burn the 7.3.8 64-bit ISO to a CD-rom (though I have heard it's possible to put the ISO onto a thumb-drive).  When you boot off the ISO, it will be like the first time you installed AMP where you run through the same exact steps that are in the Quick Start Guide (support.arubanetworks.com -> documentation -> AirWave -> 7.5.x -> Quick Start Guide).  Once this is done, you'll be good until RHEL / CentOS 7.x.  You'll be able to restore your backup after you complete the ./amp-install and go through the network setup / ssl cert key wizard.  Remember, after the install, the CLI access will be user: root, password: admin.

Did a fresh install, went through quick start, ran ./amp-install, then performed the restore... I can ping the machine and ssh into, however, the Web interface is not working.. http 500 internal server error.

Is there any errors in /var/log/httpd/access_log or error_log?  Are you logging in with a user account local on the AMP or authenticating via RADIUS / LDAP / or TACACS?  If it's the latter, try logging in as a local AMP user, navigate to AMP Setup -> Authentication tab -> reinput the passwords and then logout and try logging in again with an off-server authenticated account.  If that doesn't work, then we can try to do a web meeting and sort through this really quick.

several reboots later and she came up..  however, looking for license

I believe I have found the license file in the backup file

ok the good news, the licensing is sorted out.  However, doesnt look like the restore worked properly.  All my scan networks groups are gone.  No devices now.. ahhh

Sounds to me like the restore didn't complete all the way.  Try this to verify:

# dbc 'select count(*) from ap;'

This will tell us if the database got fully restored - as the count should be close to the device count (may have some devices that were in the new devices or ignored categories).

If it's a case that the backup didn't fully restore, then you can try doing the restore process once again.

 count
-------
     4
(1 row)

will try the restore again

[root@leviton mercury]# amp_disable
DISABLED - 1!!
[root@leviton mercury]# amp_restore -d databackup.tar.gz
databackup.tar.gz: not a file
DISABLED - 1!!
[root@leviton mercury]#

Which directory did you put the backup file in?  If it wasn't in /tmp or /root, then it might have been overwritten during the restore.

it was in /tmp, I have the file local as well.  I re-added to root with no luck.. and I moved to tmp with no luck

Restore command should be:

# amp_restore -d /tmp/databackup.tar.gz

You can verify the file exists by issuing the command:

# file /tmp/databackup.tar.gz

forgot the / before the file name.. its restoring now :)  lets hope everything restores back

seems to be finished?  do I just reboot?

Stopping postgresql service: [  OK  ]
postmaster dead but pid file exists
postmaster dead but pid file exists
postmaster dead but pid file exists
postmaster dead but pid file exists
postmaster dead but pid file exists
Starting postgresql service: [  OK  ]
Converting rrds to 64bit architecture.
Converting RRDs:   0% [                                                                   ]
Converting RRDs:  16% [===========                                            Converting RRDs:  33% [======================                                 Converting RRDs:  50% [=================================                      Converting RRDs:  66% [============================================           Converting RRDs:  83% [=======================================================Converting RRDs: 100% [=======================================                                                              DISABLED - 1!!                        
[root@leviton mercury]#
DISABLED - 1!!
[root@leviton mercury]#
DISABLED - 1!!
[root@leviton mercury]#
DISABLED - 1!!
[root@leviton mercury]#
DISABLED - 1!!
[root@leviton mercury]#

No need to reboot, just issue:

# amp_enable

Then give AMP 5-10 minutes to start up the processes.  You can issue the command:

# watch daemons

Once the list populates to more than 8 items, the WebUI should become available.

Also, try issuing the db command again to make sure all the devices were restored into the database:

# dbc 'select count(*) from ap;'

looks like we are back in business.. that restore seemed to get me back to normal.  Now can I upgrade to the latest code?

yeah, should be able to just kick off the upgrade:

# start_amp_upgrade -v 7.5.6

yep, its in progress.. thanks for all your help!

now receiving this error:  I did not perform the upgrade yet.  I started it, but then aborted it. svn: '/root/svn/mercury' is not a working copy

To fix that:

# cd /root

# ls

Do you see 2 svn directories?  If so:

# mv svn svn_broke

# mv svn_old svn

# cd /root/svn/mercury

# svn info

Confirm it's the version you were on (7.3.8)

For sanity, you may want to refresh the code to make sure it's the running copy:

# make

^^^ above command needs to be run from /root/svn/mercury ^^^

When running upgrades or backups, I suggest using the unix #screen (kb.airwave.com keyword: screen).  This will ensure that your commands run in the background should you disconnect your laptop / desktop from the network.