HP iMC High Availability

[NeilR]

Alexey,

As Justin says, SMB is used to transfer the files between the servers.. FTP on the standby is not required.

On the pimary in the DBman configuration enter the ip of the backup server and validate. If validation fails then it means the installed modules on the standby server do not match the primary. Install all the same modules and patch versions.

create a directory to hold the backups on the primary and the standby server. I beleive it must exist prior to configuring. Set that path drive:\folder on the primary and check both sets of dbs to back up and upload. Select auto-backup and set a time. Allow an hour or two to complete depending on amount of data.

On the standby select auto restore, set the path for the files, and the db's to restore. The dbman process will continuosuly scan the directory for the files and once they are transferred from the primary, it will restore them. During restore the imc processes will shutdown and restart when complete.

[pattap]

Hi Neil 

I've just started looking into setting our IMC in similar way to yours.

I'm just wondering since on standby unit files are read only what happens when your active IMC fails? If I understand correctly with HP's tool that would provide a mechanism to detect the failure standby would just take over.

What is the process in your scenario and how much of functionality on the standby unit while the active is offline you have? Is it just a case of changing the license type  from standby to active within gui to make this instance of IMC fully functional?

[NeilR]

Hi PattaP,

Some of what your get depends on what modules you are using. But the database is replicated nightly so your backup unit knows you configurations as of the last time it ran the backup. So it will continue to poll the devices it knows about and report errors based on polling, gather statistics, etc.

So you can use features like real time performance, any other performance views, real time location same as on the primary - you just can't change the configuration.

So i set my network devices to send traps to both and allow both to be accessed

I use uam to provide 802.1x and MAC authentication for all my user ports so that is one critical operation. Again i set both as authenticators on my network devices. Here IMC provides failover for the UAM processes, as the devices will interogate one and then on failure, the second one. UAM will continue to query LDAP on the standby. Some UAM processes are configurable on the standby so user sync and use managment is possible.

Alarms can be viewed but only cleared as batch not individually.

Launch of telnet, ssl and web clients to network devices is also possible.

So I just let both run in parallel. The only process I critically need to failover is UAM, and as both are active, the network device configuration handles that. 

I have not attmepted to move the active license to the standby, so can't comment there. I run these as VMs so I would likely just restore the primary and live with the loss of data if a failure so corrupted the primary. So far not been an issue.

lastly its handy for upgrades, as the upgrade processes take a while. I can do some testing by deploying on the standby first, and then if it looks good proceed to primary. Meanwhile authentication and alarming is still going on.

[pattap]

Hi @jguse 

I know this is a bit old. Quick question regarding primary/backup (stateless failover) you just described. Do both IMC instances need to be on the same IP subnet? I'd like to deploy this but ideally having both servers at two separate locations/different IP subnets. 

[jguse]

Hello @pattap 

I've usually only deployed and supported systems in this deployment model that ended up being in the same subnet, however I'm not aware of any restriction regarding this. The SMB transfer of the active system's backup to the standby system is configured by entering the IP address of the standby system in the DBMAN config, so as long as the active system can reach the standby via IP connectivity, that should work fine regardless of different subnet.