Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT cygwin DOT com Delivered-To: mailing list cygwin AT cygwin DOT com Message-ID: <3C6A581E.8ADB6088@wapme-systems.de> Date: Wed, 13 Feb 2002 13:12:14 +0100 From: Stipe Tolj Organization: Wapme Systems AG X-Mailer: Mozilla 4.7 [de]C-CCK-MCD QXW0322b (WinNT; I) X-Accept-Language: de,en MIME-Version: 1.0 To: "cygwin AT cygwin DOT com" Subject: [BUG] inter-process signaling of non-special signals breaks Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Ok, this issue has been addressed by me a couple of times, so I hope you core Cygwin developers don't get borred from me. The bug (maybe it's fixed now, but it has been detected at least from 1.3.1 up to 1.3.7) arises from my work on Apache for Cygwin. Here is the main problem scope: apache's main httpd process forkes a couple of childs that are used for handling incoming processes. Earch httpd process child listens to SIGUSR1 for gracefull shutdown. This may be used if there are too many idle childs remaining or a keep-alive (see HTTP/1.1) connection has timed out. All this seems and in fact is working! But, in some time or at some ammount of signals floating arround the SIGUSR1 signal that is passed from the main httpd scoreboard controll process to the childs is *not* received any more by the childs and hence some of them stay in keel-alive connection more forever (or at least a *very* long time). When replacing the used SIGUSR1 signal with SIGKILL everything is working as intended, even on the long term run. So my assumption is that signaling breaks at some point, either age of process or ammount of signals that the main httpd process has been issuing. When sending my hand $ kill -USR1 to one of the outstanding keep-alive childs they *do* gracefully shutdown, so I'm considering that the signal has not been coming in that has been issued by the scoreboard controlling httpd. That's all I can say from my investigations. Maybe one of the developers now known that I'm talking about and could setup a testbed for this, or at least advice me. I have looked into Cygwin signaling code, but I'm not familiar with it. Stipe tolj AT wapme-systems DOT de ------------------------------------------------------------------- Wapme Systems AG Münsterstr. 248 40470 Düsseldorf Tel: +49-211-74845-0 Fax: +49-211-74845-299 E-Mail: info AT wapme-systems DOT de Internet: http://www.wapme-systems.de ------------------------------------------------------------------- wapme.net - wherever you are -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/