undefined | Better HN

0 pointspierrefar14y ago0 comments

Hi Paul,

A site can be crawled from any number of Googlebot IP addresses, and so blocking all except one doesn't help in throttling crawling.

If you verify the site in Webmaster Tools, we have a tool you can use to set a slower crawl rate for Googlebot, regardless of which specific IP address ends up crawling the site.

Let me know if you need more help.

Edit Detailed instructions to set a custom crawl rate:

1. Verify the site in Webmaster Tools.

2. On the site's dashboard, the left hand side menu has an entry called Site Settings. Expand that and choose the Settings submenu.

3. The page there has a crawl rate setting (last one). It defaults to " Let Google determine my crawl rate (recommended)". Select "Set custom crawl rate" instead.

4. That opens up a form and choose his desired crawl rate in crawls per second.

If there is a specific problem with Googlebot, you can reach the team as follows:

1. To the right hand side of the Crawl Rate setting is a link called "Learn More". Click that to open a yellow box. 2. In the box is a link called Report a problem with Googlebot which will take you to form you can fill out with full details.

Thanks!

Pierre

0 comments

aw3c214y ago

I would like to set that crawl rate but do not see why I must register at Google to do so. Why can't Google support the Crawl-Delay directive in robots.txt for this?

Matt_Cutts14y ago

My plane is about to take off, but very briefly: people sometimes shoot themselves in the foot and get it way, way wrong. Like "crawl a single page from my website every five years" wrong.

Crawl-Delay is (in my opinion) not the best measure. We tend to talk about "hostload," which is the inverse: the number of simultaneous connections that are allowed.

anonfoobar114y ago

Another great way to shoot yourself in the foot (and getting it way way wrong) is to block all Googlebot IPs except for one.

jfoster14y ago

Instead of completely disregarding Crawl-Delay, why not support it up to a maximum value that is deemed sensible? This would prevent people from completely shooting themselves in the foot, and it would surely be better than completely disregarding it.

officemonkey14y ago

I would think that the number of people who (a) know how to create a valid robot.txt file, (b) have some idea of how to use the "crawl-delay" directive and (c) write a "shoot-themselves-in-the-foot" worthy error is vanishingly small.

3 more replies

aw3c214y ago

I do understand the thought but I think it is not a good gesture to do. You could always cap crawl-delay at a reasonable maximum and additionally allow people to fix mistakes through the webmaster tools (eg if they told your bots to stay away for a long time but in the meantime want to revert that).

Maybe instead that hostload could be parsed from robots.txt? It sure seems like the better mechanic to tweak for load issues (while traffic/bandwidth issues are still unresolved).

yesbabyyes14y ago

Matt, how does a sitemap fit into this? If I'm not mistaken, you can suggest some refresh rate there, too. Do you take that into account?

runningdogx14y ago

Could google vary the crawling rate on each site and see what effects that has on response times, and develop an algorithm to adjust crawl speed so as not to affect site performance too much? If google starts crawling a site and notices sequential crawl requests are answered in .5s w/ .1s stddev and it starts crawling with 10 parallel connections and the answers are 2s w/ 1s stddev, clearly that's a problem because user experience for real people will be impacted. Maybe google could automatically email webmaster@ and notify them of performance issues it sees when crawling.

Another thing that might help google is for them to announce and support some meta tag that would allow site owners (or web app devs) to declare how likely a page is to change in the future. Google could store that with the page metadata and when crawling a site for updates, particularly when rate limited via webmaster tools, it could first crawl those pages most likely to have changed. Forum/discussion sites could add the meta tags to older threads (particularly once they're no longer open for comments) announcing to google that those thread pages are unlikely to change in the future. For sites with lots of old threads (or lots of pages generated from data stored in a DB and not all of which can be cached), that sort of feature would help the site during google crawls and would help google keep more recent pages up to date without crawling entire sites.

ricardobeat14y ago

> declare how likely a page is to change in the future

I believe you can do that using a sitemap.xml

j / k navigate · click thread line to collapse

0 pointspierrefar14y ago0 comments

Hi Paul,

A site can be crawled from any number of Googlebot IP addresses, and so blocking all except one doesn't help in throttling crawling.

If you verify the site in Webmaster Tools, we have a tool you can use to set a slower crawl rate for Googlebot, regardless of which specific IP address ends up crawling the site.

Let me know if you need more help.

Edit Detailed instructions to set a custom crawl rate:

1. Verify the site in Webmaster Tools.

2. On the site's dashboard, the left hand side menu has an entry called Site Settings. Expand that and choose the Settings submenu.

3. The page there has a crawl rate setting (last one). It defaults to " Let Google determine my crawl rate (recommended)". Select "Set custom crawl rate" instead.

4. That opens up a form and choose his desired crawl rate in crawls per second.

If there is a specific problem with Googlebot, you can reach the team as follows:

Thanks!

Pierre

0 comments

aw3c214y ago

I would like to set that crawl rate but do not see why I must register at Google to do so. Why can't Google support the Crawl-Delay directive in robots.txt for this?

Matt_Cutts14y ago

My plane is about to take off, but very briefly: people sometimes shoot themselves in the foot and get it way, way wrong. Like "crawl a single page from my website every five years" wrong.

Crawl-Delay is (in my opinion) not the best measure. We tend to talk about "hostload," which is the inverse: the number of simultaneous connections that are allowed.

anonfoobar114y ago

Another great way to shoot yourself in the foot (and getting it way way wrong) is to block all Googlebot IPs except for one.

jfoster14y ago

officemonkey14y ago

3 more replies

aw3c214y ago

Maybe instead that hostload could be parsed from robots.txt? It sure seems like the better mechanic to tweak for load issues (while traffic/bandwidth issues are still unresolved).

yesbabyyes14y ago

Matt, how does a sitemap fit into this? If I'm not mistaken, you can suggest some refresh rate there, too. Do you take that into account?

runningdogx14y ago

ricardobeat14y ago

> declare how likely a page is to change in the future

I believe you can do that using a sitemap.xml

j / k navigate · click thread line to collapse