--- "Richard L. Hamilton" <[EMAIL PROTECTED]> wrote:

> > 
> > > "What, Slowlaris still exists?"
> > 
> > Well, I sure was blown away by the incredible
> > difference between Solaris 8 find and gnu find.
> > 
> 
> This would be the gnu find that needs one to use a
> nonstandard option
> -noleaf to tell it _not_ to optimize out one stat()
> per directory?  That
> could've been done automatically by testing the
> assumption on each
> filesystem, and altering its behavior adaptively,
> but no, they had to
> add an extra flag to ensure it behaves correctly on
> AFS and ISO filesystems.

No idea. Maybe you can tell me what it was. A qmail
queue on a UFS filesystem with somewhere around 500k
mails in the queue. A script that checks the queue
size basically runs something like this: 'find
queue/mess/* -print | wc -w'

Solaris find took over 24 hours and still I got no
result and was waiting. After I asked my manager why
does it take so long, he told me to give gnu find a
shot. In seconds I got the answer. This gave me two
impressions. 1) The solaris kernel had great I/O.
Using just seconds to report about 500 thousand files
is something I do not remember with 2.4 linux distros
even when there is no i/o contention. 2) Solaris find
sucked.

maybe you can tell me what is what?

> 
> Performance gained at the expense of incorrect
> behavior that requires
> special usage to avoid strikes me as highly suspect.

Please explain.

> 
> (OTOH, for all I know it has other features and
> optimizations that are
> perfectly reasonable; and as long as they aren't
> incompatible, I wouldn't
> have a problem with them being adopted.)

... so do you know the differences or not?

Send instant messages to your online friends http://uk.messenger.yahoo.com 
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to