This is a post about my struggles with prerender redirects using iron router. I’m not sure if this problem is solvable, I may have to go back to the drawing board. Maybe someone here may have similar experiences.
So I have a thing in my app where if you type
www.example.com/todo/123456789
(123456789 would be an id)
You’ll get redirected to a more descriptive URL
www.example.com/todo/123456789-buy-eggs
Even something like
www.example.com/todo/12345678-fijsdfoijdio
Would also redirect to
www.example.com/todo/123456789-buy-eggs
There is one single canonical URL for each todo item, but potentially infinite redirects. For google I use a canonical URL meta tag, but that’s irrelevant to prerender.
https://prerender.io/documentation/best-practices
So I’m trying to prerender this route. And using prerender best practices, I
- Turn off this redirect if the user-agent is prerender. (So the url stays www.example.com/todo/123456789)
- I insert into the meta tags
prerender-status-code -> 301
prerender-header -> The canonical URL (Location: www.example.com/todo/123456789-buy-eggs). - I set window.prerenderReady = true
So the expected behavior would be that prerender would only cache the canonical page, and redirect these other potentially infinite URL’s to that cached page when accessed by googlebot.
But that’s not happening.
(Sidenote: All the cases I’m about to go over happen before window.prerenderReady is set to true, I’ve determined that’s largely unimpactful).
First I tried setting the meta tags in the blaze template onRendered This didn’t seem to work.
Next I put them in the routers onBeforeAction. That also didn’t work.
Finally I put them in the routers waitOn, doing it immediately after the publication for todos return (Which is necessary to get the canonical URL). This also did not work.
What did end up working best was actually my first idea, which was using ostrio:iron-router-meta to set the router tags that way. When I did this, the expect behaviour that I want (Prerender sensing a 301 and delivering the redirected page) happens.
But ONLY when I spoof googlebot myself in chromes dev tools and visit my website.
Googlebot itself does not see a difference. It still makes requests that prerender doesn’t sense the 301 for and makes false dupe pages still when it asks for them, despite it being the same page I’ve tested myself when spoofing googlebot and successfully getting the 301 code in prerenders console and correct behavior that I want.
So I’m scratching my head, the only difference I’ve noticed with the meta functions from ostrio:iron-router-meta is that they run more then once. I suspect it’s because they’re reacting to publication data coming back from the server.
So you might think the problem is that they’re being filled with bad data because the mongodb collections aren’t initially available yet to make the canonical URL off of. Except I’m using fastrender which solves that problem and they seem to run correctly even the first time they’re called.
So I don’t know.
Another interesting note is that marking a page as “404” is trivial. I can do it on the onRendered callback even (Which is the final piece of code that runs before prerenderReady is set to true) and it works just fine for me OR the actual googlebot. I wonder if there’s just some oversight I’m missing.
