httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From d...@ast.cam.ac.uk (David Robinson)
Subject Re: Nameserver cache
Date Thu, 23 Mar 1995 13:20:00 GMT
Wow, what a lot of messages. I'll try and deal with them all here.

1. I have uploaded a new version of the patch that should be more portable.
   Please try it again if you had problems with the previous version.

2. For the particular case of SunOS 4, it appears that adding -lresolv to
   the link command circumvents NIS. This should be documented in the Makefile.

3. I _need_ this patch, or some other change to httpd to avoid NIS+.
   Altering nsswitch.conf is not an option. (We run Solaris 2.4 here.)
   I would be quite happy to try out patches to httpd which used res_mkquery
   and friends, as an alternative. 

4. I have no problem with increasing the cache size, as long as this can
   be justified.

5. I have not been sufficiently precise in my description of how the cache
   works. It does NOT use pipes for a client/server cache access strategy (in
   the manner of in.named & sockets). The cache is in normal memory that is
   duplicated by fork() from the parent to the child processes. To search
   the cache, httpd merely runs through an array.
   The pipe is only used for cache misses, where the child process sends
   new data to be entered in the cache back to the parent. This is an
   asynchronous message that does not require a response.
   This approach is lightweight, and I would expect it to be significantly
   faster than talking to a named via UDP or TCP. But by all means please
   test that assertion.


 David.

Mime
View raw message