Ability to keep indexing beyond cron failures

In many installations, "Last cron did not complete" messages start to appear in the logs, and all search indexing stops at a specific number of remaining nodes. The problem is that if a cron does not complete with a task of indexing 10 nodes, and does only 9, those 9 are not recorded to have been indexed, and the next cron falls in the same trap, thus stopping all indexing. Crons probably stop because of a timeout, although other causes may exist.

We can record each indexing operation by calling node_update_shutdown() at the end (and inside) of the while loop in node_update_index() of the node module.

I have done it many times and it save the situation; cron runs still do not complete, yet the indexing progresses with each cron run. I recommend this hack for anyone with "cron did not complete" messages!

This applies to drupal 4.x and 5.x.