Robert_Charlton - 8:02 pm on May 20, 2009 (gmt 0)

I may be misinterpreting what you've done, but this comment jumped out at me....

We set a meta robots on each https page rather than the root level robots.txt

I'm not quite understanding how you separated https pages from http pages to do this... and particularly whether you put the meta robots tag just on the pages that should be secure (like login and checkout pages), or whether you're talking about product pages that inadvertently became indexed as https.

On the "900 https site product and category pages" you mention, the pages are the same. What happens when you have an https canonical issue is that the same pages are seen under multiple urls, and the issue is one of references to the http pages that are causing them to be indexed also as https, combined with a server setup that hasn't canonicalized these pages as http.

The 301 redirect, if properly done, should stop the https indexing problem... but, if you've also put the meta robots tag on all of those 900 pages, you're likely to have another problem....

My emphasis...

Now that the indexed https pages have dropped out of the index, does anyone know how long it might take google to start listing the old http pages once again?

If you've got meta robots noindex on all those pages, Google won't reindex them. It will drop them all, regardless of whether they're https or http.

Again, I may be misinterpreting what you've done, but the above is a possible scenario that occurred to me.