Rogerbot getting cheeky?

[BM7]

Hi SeoMoz,

From time to time my server crashes during Rogerbot's crawling escapades, even though I have a robots.txt file with a crawl-delay 10, now just increased to 20.

I looked at the Apache log and noticed Roger hitting me from from 4 different addresses 216.244.72.3, 72.11, 72.12 and 216.176.191.201, and most times whilst on each separate address, it was 10 seconds apart, ALL 4 addresses would hit 4 different pages simultaneously (example 2). At other times, it wasn't respecting robots.txt at all (see example 1 below).

I wouldn't call this situation 'respecting the crawl-delay' entry in robots.txt as other question answered here by you have stated. 4 simultaneous page requests within 1 sec from Rogerbot is not what should be happening IMHO.

example 1
216.244.72.12 - - [05/Sep/2012:15:54:27 +1000] "GET /store/product-info.php?mypage1.html" 200 77813
216.244.72.12 - - [05/Sep/2012:15:54:27 +1000] "GET /store/product-info.php?mypage2.html HTTP/1.1" 200 74058
216.244.72.12 - - [05/Sep/2012:15:54:28 +1000] "GET /store/product-info.php?mypage3.html HTTP/1.1" 200 69772
216.244.72.12 - - [05/Sep/2012:15:54:37 +1000] "GET /store/product-info.php?mypage4.html HTTP/1.1" 200 82441

example 2
216.244.72.12 - - [05/Sep/2012:15:46:15 +1000] "GET /store/mypage1.html HTTP/1.1" 200 70209
216.244.72.11 - - [05/Sep/2012:15:46:15 +1000] "GET /store/mypage2.html HTTP/1.1" 200 82384
216.244.72.12 - - [05/Sep/2012:15:46:15 +1000] "GET /store/mypage3.html HTTP/1.1" 200 83683
216.244.72.3 - - [05/Sep/2012:15:46:15 +1000] "GET /store/mypage4.html HTTP/1.1" 200 82431
216.244.72.3 - - [05/Sep/2012:15:46:16 +1000] "GET /store/mypage5.html HTTP/1.1" 200 82855
216.176.191.201 - - [05/Sep/2012:15:46:26 +1000] "GET /store/mypage6.html HTTP/1.1" 200 75659

Please advise.

[MeganSingley]

Hi there,

This is Megan from the SEOmoz Help Team.  I'm so sorry Rogerbot is causing you grief!  This actually might be happening because your crawl delay is too long, so rogerbot just ends up ignoring it so he can complete the crawl.  If you set your crawl delay to a max of 7, then it should solve your problem.  If you're still running into issues, though, please send us a message to help@seomoz.org and we'll check it out asap!

Cheers!

[BM7]

Thanks Megan for your reply,

Will give that a try and have blocked 2 addresses so you are reduced to 2 crawler sessions. These two measures should reduce the load considerably as long as Rogerbot respects the 7 second delay.

IMHO ignoring the Crawl-Delay set by the webmaster of the site you are crawling, which crawlers are supposed to respect, is wrong. I got a Google WMT nasty for being down 5 hours due to Rogerbot as it was the middle of the night so only got restarted in the morning.

Also, my site has around 600 discrete pages of which you crawl about 500, so even at the original 10 seconds crawl delay you could do my whole site in less than 1.5 hours, which is only required once a week. So in my mind that suggests there is no need to overrule my settings in robots.txt 'so he (Roger) can complete the crawl'.

Regards,

[MeganSingley]

Hi BM7,

I'm going to open up a ticket on this to have our engineers take a closer look at your site.  Once we have an overall response, I'll post it here for other community members to view.  

Cheers!