pinch — 2011-04-05T14:05:07-04:00 — #1
I need to take down my site for awhile but don't want the web crawlers to think the pages are unavailable. I normally use app_offline.htm to bring down my site but I'm seeing conflicting information as to what HTTP Status Code is presented. My understanding is that you should use status code 503 to tell the crawlers the page is temporarily unavailable, but some resources say that this results in a 404 status code to be presented?
wwb_99 — 2011-04-05T14:45:16-04:00 — #2
How long? Hours? Days? Weeks?
pinch — 2011-04-05T14:53:54-04:00 — #3
Let's say 2 days.
wwb_99 — 2011-04-05T15:36:33-04:00 — #4
I'd do something a bit fancier than app_offline.htm that returns the right status codes then -- that is for a quick fix when you want to stop things from banging the app while you update, not for extended downtime.
Really I'd find a way to avoid extended downtime, but I've spent too long in operations . . .
pinch — 2011-04-05T21:12:38-04:00 — #5
That makes sense and I agree, I just didn't know if there was a way to extend downtime and still not be penalized by the SEs. Thanks.
pinch — 2011-04-06T08:49:08-04:00 — #6
So are you saying that app_offline.htm returns the correct 503 HTTP status code? Also, how long do you think it is appropraite to have a site down for upgrade using app_offline.htm?
wwb_99 — 2011-04-06T09:54:02-04:00 — #7
Honestly no idea what code -- really don't use it, we swap versions so our site might blip for 5 seconds. As for extended downtime, I'm generally in the school of thought that you shouldn't have downtime longer than that in the modern era.
pinch — 2011-04-06T10:04:22-04:00 — #8
I hear ya. Normally my upgrades don't take that long but this one involves wholesale DB modifications and blog integration which will take awhile.
What do you mean by swap versions?
wwb_99 — 2011-04-06T10:52:51-04:00 — #9
Could be lots of things. First, those sorts of things happen in QA/staging here. By the time we hit production, it is really a matter of svn switch + run the database update script. Some crazier things have involved swapping out entire VMs. And we've started using some http reverse proxies so swapping can then become switching what the proxy is forwarding to. Meaning upgrades are completely non destructive.
pinch — 2011-04-06T12:57:26-04:00 — #10
We do a similar thing here at work but unfortunately I'm talking about a personal project. Thanks for the advice.