I'm not sure how the httpd processes are being tied up, but "tied up" imapds that are otherwise idle don't cost you anything except some swap and a process table entry. They're basicly free.
Unless the imap connection has a mailbox open. That ties up a potentially very large memory resource (depending on the size of the mailbox) on the server.
I have to say I don't know for certain whether that's true for the cyrus imapd, but for other imapds I've used, that has been the case. Someone know for sure?
Ted Fines
--On Wednesday, June 11, 2003 2:03 AM -0400 Rob Siemborski <[EMAIL PROTECTED]> wrote:
On Tue, 10 Jun 2003, Gary Mills wrote:
We do use that, and it probably does improve performance. It does have a problem with idle browser connections that accumulate with time. This also ties up a lot of `imapd' and `httpd' processes. It probably needs a client timeout someplace. I haven't had time to investigate further.
I'm not sure how the httpd processes are being tied up, but "tied up" imapds that are otherwise idle don't cost you anything except some swap and a process table entry. They're basicly free.
-Rob
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456 Research Systems Programmer * /usr/contributed Gatekeeper
