Automic High Availability

[Genarro Montano]

I am working with Automic v12.3 and we are looking at establishing HA for our environment. During our installation, we configured our ucsrv.ini file to include seperate ranges for cps and wps. Primary is using cp ports 2217-2221 and secondary is using 2222-2226. Similarly, we used the same settings for our WPs. Now, wouldnt both Engines be running using different CPs and WPs? I am wondering what steps we must take to successfully establish High Availability? As of now we cannot start our Secondary Engine Processes so I know we are missing something just not entirely sure what.

[Frank Muffke]

Hi

from V12.3 you do not need to define WP Ports in ucsrv.ini file (exception PWP).

I only defined PWP (must be unique) , CP.PORTS range, JCP Port and JWP.SYNC.PORTS

We have 2 linux machines where AE runs on it.

The more tricky is to get the DB high available - usually (always)  DB is the bottleneck :-)

From Automic View you need to provide a clustered DB - we have an oracle RAC that is somehow split up to 2 more servers on 2 locations.
But don' t ask me how exactly this is done, this DB config stuff is "devil's work" :-)

cheers, Wolfgang

------------------------------
I know I do really know it!
------------------------------

[Carsten Schmitz]

(edit: clarified the manual start procedure for debugging java processes)

As far as I know, much of the coordination for the server processes on two distinct servers works via the database. Server processes on both servers connect to the db, and get info about their partner servers from there.

Once you get your processes up on the second server, you should ideally have workers and cp (and jwp and jcp) running on both servers, but only ever have one PWP. If the PWP dies, another worker should take over the role. If server #1 dies, a worker on server #2 should take over the PWP role. The PWP always grabs the port configured in the ini file, and PWP election usually takes under a minute for us.

As to why your processes on server #2 aren't coming up: double-check the ini file - as @Frank Muffke said, use the new ini file template from 12.3 and move your old values across if needed, to not copy an old ini file into 12.3 because the syntax changed. Check the logs in the Automic(!) "tmp" directory. Also check the smgr logs (in smgr/tmp) for any startup problems. If you are using Linux and systemd, and the problem is with Java processes, also try starting the processes manually by just running the "java -jar whatever" command as found in your smgr's .smd file on the console. Otherwise you might miss error messages going straight into nothingness in the usual smgr setup. The very last resort then, failing all of the above, is "strace".

Hth,

[Genarro Montano]

So each Engine (both Primary and Secondary) would need their own PWP? or are you saying that it is just one unique one that they both share?
And yes I agree about that DB nonsense!

[Carsten Schmitz]

One PWP per installation. Both servers share the one unique PWP, where ever it may be running :)

[Naresh Sunkar]

Hi ,

is there any steps to do the PWP configuration , kindly share the URL

[Genarro Montano]

Thank you both for the information!