Upgrade considerations for a Highly Available environment
There are considerations to be aware of when upgrading within a Highly Available (HA) environment.
For the most part, upgrading an HA HCL Traveler server is the same as upgrading a stand alone HCL Traveler server, but there are a few things to keep in mind. These items are outlined in the following list. Be sure to also review the general Upgrade considerations and overview before performing any upgrade.
- The correct way to upgrade an HA pool is to upgrade one server at a time. Complete the upgrade fully on one system, including an HCL Domino® Server upgrade if needed, and bring the server back online before proceeding to the next system. Performing the upgrade in this manner ensures continued service during the upgrade.
- It is recommended that all servers in an HA pool be upgraded within a short amount of time. That is, do not upgrade one server in the pool, then run it for an extended period of time. Having the HCL Traveler servers at different levels affects the load balancing algorithms. In the short term, this is no concern, but over a span of time, such as several days, the load balance algorithm may begin to overload the newer servers instead of the older ones. This is done intentionally to ensure the best user experience. For example, if a user is able to take advantage of a new feature, that user will no longer sync against an older server where the new feature is not available.
- If you are using Auto Schema updates, the first server to be upgraded will update the database
schema as necessary. This will not effect the existing older servers. However, it will cause the
startup process to take slightly longer than normal for the first server. Verify the server is up
and running before bringing down the next server in the pool. One of the easiest ways to verify a
server is functioning is to log in directly to the HCL Traveler servlet on that system, generally
https://hostname/traveler
. If the server is functional, you will see user and server status messages display. - If you are not using Auto Schema updates (for example, if you
have
NTS_AUTO_DBSCHEMA=false
set in thenotes.ini
, then you will need to manually update the database schema on the enterprise database system. You should do this as part of the upgrade of the first system in the HA pool. See Updating the enterprise database for more information. - If for some reason it is not possible to upgrade a system in the pool, then set up and add a new system to the HA Pool, then retire the old system. All Traveler servers in the pool must be at the same Traveler version, but they do not need to be at the same OS level or even Domino® level, although it is recommended when possible. See Moving HCL Traveler to a new server for more information.
- Due to an issue with the handling of different server levels between the HCL Traveler servlet running in the HTTP task and the HCL Traveler task itself, HCL Traveler cannot have a mix of 8.X and 10.X Traveler servers in an HA pool as they cannot communicate with each other between HTTP and Traveler tasks. If you are upgrading an HA pool from 8.X to 10.X, you need to do all of the servers as quickly as possible (either all at once by taking an outage or incrementally but as quickly as possible) to avoid this cross communication issue. This is only a problem when upgrading an 8.x HA Pool, upgrading any 9.0x or later environment does not have this limitation.
- It is recommended that all Domino servers in a HCL Traveler HA pool be at the same level. For example, it is not recommended for them to run one on an HCL Traveler server on Domino 10.0.0 and another on an HCL Traveler server on Domino 11.0.0 in the same pool. During an upgrade this scenario may be unavoidable but it is recommended that you minimize the length of time you run with mixed versions. The same applies with Domino Hot Fixes and Fix Packs, as some fixes are critical to HCL Traveler functionality.