Question

Does Worklight supports server upgrade (e.g. apply fix pack, upgrade minor version, upgrade major version) without server downtime in cluster?

My understanding is that for server upgrade, there are some scripts being applied in the database. Moreover, there are some tasks being executed to sync between database and the file system.

Is it possible to upgrade the servers one by one, or sync it after one has been updated without any downtime? If so, could you please provide me the procedures? Thanks!

Environment Info: Worklight Consumer Edition 5.0.6.1, WAS 8.5, Oracle 11g RAC

Était-ce utile?

La solution 2

The below may not be correct, depending on your upgrade scenario:


In general: it is not yet possible to upgrade the servers in a cluster one-by-one. You will need to shut down all servers in the cluster and upgrade them all.

That said, I don't know what is your network topology... and an upgrade scenario can and will change depending on that.

It could be that you have 2 clusters of Worklight Servers, each with its own database, and a load balancer that redirects app requests:

[load          balancer]
   |              | 
   |              |
   |              | 
[WL cluster]   [WL cluster]
-s -s -s -s    -s -s -s -s
   |              |
  [DB]           [DB]

In such a case you can:

  1. Direct all traffic to one cluster
  2. Take down the other cluster, upgrade all servers, restart
  3. Direct all traffic to the upgraded cluster
  4. Take down the unupgraded cluster, upgrade all servers, restart
  5. Direct again the traffic to both clusters

Autres conseils

The documented in place upgrade procedures (http://pic.dhe.ibm.com/infocenter/wrklight/v6r1m0/topic/com.ibm.worklight.upgrade.doc/topics/t_upgrading.html) include some downtime.

Issues to anticipate to minimize downtime include: - porting the server side artifacts (wlapp and adaptaers) to the new version and being ready to upload them (not needed in 6.1 that provides separation of studio and server lifecycle) - database schema update

In the 6.1 version the downtime for an in-place upgrade can be minimized to this:

  1. stop servers
  2. upgrade database
  3. upgrade the first server and restart it.

-> If Application Center is not installed or not installed on the same application server than the worklight server.

This is summarized in the table attached to this page: http://pic.dhe.ibm.com/infocenter/wrklight/v6r1m0/topic/com.ibm.worklight.upgrade.doc/devenv/c_upgrade_to_srvr610_overview.html

Licencié sous: CC-BY-SA avec attribution
Non affilié à StackOverflow
scroll top