From owner-freebsd-hackers Fri Jun 16 09:30:31 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id JAA24477 for hackers-outgoing; Fri, 16 Jun 1995 09:30:31 -0700 Received: from aries.ibms.sinica.edu.tw ([140.109.40.248]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id JAA24469 for ; Fri, 16 Jun 1995 09:30:24 -0700 Received: (from taob@localhost) by aries.ibms.sinica.edu.tw (8.6.11/8.6.9) id AAA00331; Sat, 17 Jun 1995 00:27:51 +0800 Date: Sat, 17 Jun 1995 00:27:50 +0800 (CST) From: Brian Tao To: "Julian Stacey " cc: FREEBSD-HACKERS-L Subject: Re: Too many open files in system In-Reply-To: <199506152003.WAA05835@vector.eikon.e-technik.tu-muenchen.de> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: hackers-owner@freebsd.org Precedence: bulk On Thu, 15 Jun 1995, Julian Stacey wrote: > > Maybe something other than httpd is consuming your file descriptor table ? But what? This can happen on a freshly-rebooted machine and practically nothing else running except the http server. The process table looks something like this: PID TT STAT TIME COMMAND 0 ?? DLs 0:00.00 (swapper) 1 ?? Is 0:00.07 /sbin/init -- 2 ?? DL 0:02.29 (pagedaemon) 3 ?? DL 0:00.92 (vmdaemon) 4 ?? DL 0:00.78 (update) 61 ?? Ss 0:00.23 syslogd 71 ?? IWs 0:00.02 portmap 81 ?? IWs 0:00.29 inetd 164 ?? Is 0:00.05 httpd-root (apache) 165 ?? I 0:00.01 apache-0- (apache) 166 ?? I 0:00.01 apache-1- (apache) 167 ?? I 0:00.01 apache-2- (apache) 168 ?? I 0:00.01 apache-3- (apache) 169 ?? I 0:00.01 apache-4- (apache) 170 ?? I 0:00.01 apache-5- (apache) 171 ?? I 0:00.01 apache-6- (apache) 172 ?? I 0:00.02 apache-7- (apache) 173 ?? I 0:00.01 apache-8- (apache) 174 ?? I 0:00.01 apache-9- (apache) 322 v1 R+ 0:00.01 ps ax 213 v1 IWs+ 0:00.47 -csh (tcsh) 146 v2 IWs+ 0:00.04 /usr/libexec/getty Pc ttyv2 147 v3 IWs+ 0:00.04 /usr/libexec/getty Pc ttyv3 148 v4 IWs+ 0:00.04 /usr/libexec/getty Pc ttyv4 149 v5 IWs+ 0:00.04 /usr/libexec/getty Pc ttyv5 150 v6 IWs+ 0:00.04 /usr/libexec/getty Pc ttyv6 That's a pretty bare system. Yes, I did try it with latest Apache server that supports the pre-fork model (with 50 children, not just the ten shown here). It died about 20 minutes into the test. Same symptoms as before: all swapping suddenly ceases and the machine "sort of" locks up. It's hard to describe. Telnets to other machines still work just fine, and I can hit Enter on a local shell and get the shell prompt back, or I can continue to read mail in Pine. But I cannot start a new process. Typing in a command in the shell just hangs there (as if my NFS disks disappeared, but they didn't, since I don't see any syslog messages about that). Running a shell command from vi also hangs. Trying to login on another virtual console hangs after I enter the username (no Password prompt). Very strange. I'm running it now with just 10 servers (and clients) to see if it can make it through the night this way. -- Brian ("Though this be madness, yet there is method in't") Tao taob@gate.sinica.edu.tw <-- work ........ play --> taob@io.org