On Thu, Aug 30, 2001, Joe Cooper wrote:
> So far so good. Except for one site that gets mangled about 50% of the
> time. I've been trying to reproduce the problem elsewhere, because the
> site is an authenticated site (non-SSL). I can't seem to reproduce the
> problem on any other site--but it never happens without the proxy. Very
> strange. I'm going to try to isolate the problem to either the size of
> the page (which is a huge page filled with tables) or the authentication.
Weird.
> Very fast though. COSS might just give reiser_raw something to sweat
> about (of course, since reiser_raw is dead or at least comatose without
> tons of funding, I guess it won't be sweating too much). I won't know
> for sure until I actually put it to a full set of benches including a
> polymix-[34] load and a long term set of other tests, but it looks good
> so far.
Right. Well, the reiserfs_raw code may be stale but I'm planning on
using it as one of the new-style FSes once I've cut some new HTTP code.
Its certainly the eaiest FS to *use* for this. :-)
I'm also interested in general stability rather than speed.
The diskd/ufs/aufs code should still Just Work(tm).
I'm going to be reworking quite a bit of the storage read/write path
along with a few other storage manager "optimisations" that may
or may not result in a slight speed increase.
I really need to throw in the >2gb store option one of these days
to make it useful by normal people.
Adrian
Received on Fri Aug 31 2001 - 04:46:20 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:14:17 MST