Questions about rebuilding the search index from command line (via searchindex.php)

[Silviu]

Hello everyone,

Last week we took the major leap of upgrading from vB 3.8 to vB 4.1.3, and while the update went fairly smoothly, there's one major "hiccup" that we haven't solved: the search function.

We tried Sphinx Search (thanks to the plugin created by Digital Point), but despite the obvious improvement in speed, Sphinx is incapable (at the moment) of mimicking the old sort style (by the date of the last reply in a thread), which has our users very confused, especially when they're using the Quick Search function or looking for the replies of other users. So, for now, our only option is to go back to the default vB 4 search. Which brings us to the current situation:

Since our post table has close to 9 million entries, we are going to rebuild the index from the command line using the searchindex.php file. And while we did a test run on a development server to check out the options, before we set the world on fire we'd like to know the following:

1. How big of a difference is it between the "Normal" and "Agressive" modes in terms on speed and is it worth the risk?

The php file itself states that:

Indexing posts can be done in 2 ways: By using the built-in indexer functions (normal) or by bypassing them completely and using SQL commands only (agressive). Agressive mode is usually a lot faster but it does not guarantee 100% correct results.

2. Once we start the script, can it be stopped with a simple CTRL + C from the bash prompt, or does it pass on everything to the SQL server and the only way to stop it is to restart the SQL server?

3. How does the "Turn off DB indexes during rebuild" option impact the rebuild? Should it be on or off (we intend to keep our forum live during the procedure)?

4. Can the script be run in batches? Say, index 1.000.000 posts today, stop the script, backup the DB, index the next million posts tomorrow from where it left off and so on?

5. Just making sure, when the script asks for input (for instance Enable Agressive Mode [0/1=Default=0]), we're assuming 0 is for no and 1 for yes, correct?

[charlesr]

I'd like to know the answers to this too please.