During one of my recent MOSS projects, we have found out that Index Server (with MOSS RTM) in the farm wasn’t up to the same patch level as WFE Server (with Infrastructure Update) hosting central admin site. During the stabilization phase prior to SP1 upgrade, we wanted to upgrade the index server to the same level as the WFE server.
You cannot upgrade or install the patch on the server if it’s different than central admin server patch level in the running farm due to database version mismatch with the server binaries. In this case, infrastructure update installation media would fail. To resolve the issues, you must remove the server from the farm, rebuild the server with the required patch level, and reattach to the farm.
To move the Index Server from one server to the another server, typical steps would involve –
- Stop the MOSS Search Service from the server if it’s running and start the MOSS Search Service on the new server.
- Update the index server from the SSP Configuration page from the SSP Admin page.
- Remove the index server from the farm either by uninstalling the MOSS binaries from the server (preferable method if you are going to rebuild the server) or remove the server from the servers in the farm page of the central admin site (preferable method if you no longer has server available or you want to keep the binaries on the server to reuse it for other purposes in the farm).
In most cases, moving the index service host role from the one server to the other server, should run smoothly but if you notice that search service status is other than idle (e.g. retrieving or offline) on the search services page on the central admin site applications tab or search seems not working properly or accessing search settings page from the SSP admin site times out, there must be something wrong with the MOSS index reset process.
To resolve the many issues (Either errors in the event log or seemingly no errors in the event log) related to moving the index server, you might want to try following steps to fix the problems. If this doesn’t fix the problems then you might want to research further.
- Stop and Restart the Office Search role. You can use stsadm or central administration web site to perform the tasks. Remember central administration site creates a timer job for each requests. It might perform little bit slower than stsadm. Use the STSADM to stop the service – stsadm -o osearch -action stop and start the service – stsadm -o osearch -action start -role IndexQuery -farmserviceaccount DOMAIN\accountname -farmservicepassword PASSWORD
- Edit the SSP page and select the new server as a index server.
- Reboot the index server. This is very important step. During moving the index role process, sometimes farm gets stuck and rebooting the server might free up the RAM for the index gatherer service.
- Visit the Manage Search Settings page and hopefully crawling and propagation status is no longer retrieving and idle.
- Visit Shared Services Provider Site, Search Settings page, and Restart the full crawls of the content sources