This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug network/22514] resolver don't iterate all search domains when reaching timeout


https://sourceware.org/bugzilla/show_bug.cgi?id=22514

--- Comment #2 from i at f2light dot com ---
(In reply to Florian Weimer from comment #1)
> I think it is very questionable to switch to a different query domain (along
> the search path) if you never get a NXDOMAIN/NODATA reply because as a
> result, (temporary) network glitches alter name resolution results.
> 

Agreed! That's why I got really confused when resolver didn't stop after the
first domain's lookup failure but start looking up a domain without search
domains. Laterly I decided to report it.

For offline machines, that wastes half of my waiting time.
For online machines, a temporary network failure could give me a different
answer (may be somehow a little security related?).

>From my personal aspect, I'd prefer returning EAI_AGAIN immediately whenever
reaching timeout. (Don't query any other domains)

> On the other hand, there might be setups out there which rely on the current
> behavior to some extent.

That depends on how you regard this. :)
A bug must be fixed, as I can't say "Don't fix dirtycow because my program
requires that to work".
However, if that's a feature, I do accept it. Just say "we code that intended".

Anyway, making an option to offer different behaviors does work for
compatibility, if necessary.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]