Comment #36 on issue 12754 by ben.rowland: "Resolving host" takes forever;  
disabling prefetching helps
http://code.google.com/p/chromium/issues/detail?id=12754

@jar:
Thanks for your comments.  I haven't got to the bottom of this yet, but  
more results
are in.

I have www.google.com as my homepage, which seems to redirect to  
www.google.co.uk at
some point.  The histograms I sent were collected immediately upon startup,  
as soon as
the problem was first observed.

I am away from home at the moment, but will try the same tests against  
google's public
DNS servers when I get back.

I tried running nslookup against www.google.com a number of times, and  
could not
reproduce the delay.  I'm not sure where the resolution is being done, but  
I guess by
my ISP.

I recompiled chrome with some printfs, and isolated the delay to be within  
the call to
the syscall getaddrinfo(), in host_resolver_proc.cc.  So it's probably  
happening
outside of chrome, excepting any strange thread-scheduling issues which  
seem unlikely.
The delays I recorded through this method follow an interesting pattern.   
It always
seems to be either ~30ms (normal), or a multiple of 5 seconds + ~30ms, so  
e.g. 5037,
10027, 15045ms.  It varies up and down, but gives me a feeling that the  
delay gets
worse as the load piles up.  This looks to me like a well-defined 5-second  
timeout is
happening somewhere, rather than just general slowness.

I wrote a little test program which recreates the call to getaddrinfo  
(where the
slowness was seen) with the same parameters.  But the problem hostnames was  
resolved in
an instant.  Wish it were as simple as that.  I'll try some load-testing  
once I get
back to recreate the conditions of the prefetching on chrome startup - if I  
can
reproduce it outside of chrome that would be interesting.

The fact that this problem only appears during prefetching is interesting -  
it's as if
it's the heavy load of rapid DNS requests which triggers it.  If it turns  
out to be a
problem with my ISP or LAN, it would be useful for other chrome users to  
know the
resolution, as prefetching seems to set up some unique conditions.

--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings

-- 
Automated mail from issue updates at http://crbug.com/
Subscription options: http://groups.google.com/group/chromium-bugs

Reply via email to