This is a copy of a email a wrote earlier to the author of xinetd. It contains all relevant info: ---------------------------------------------- Hi, Are you aware of any problem with xinetd that causes it to "forget" processes it has spawned ? I run a rather busy FTP-server (vsftpd-0.9.3) on a Red Hat 7.2 machine via xinetd-2.3.3. I configured a "instances=200" limit for the ftp service, but I notice that there are about 300 processes running now and the number keeps increasing. I've made a dump via a "kill -USR2" and it shows this (older dump): ====================================================================== Near the top: ------------- Service = ftp State = Active Service configuration: ftp id = ftp socket_type = stream Protocol (name,number) = (tcp,6) Instances = 200 Nice = 10 Groups = 0 Server = /usr/sbin/vsftpd Server argv = vsftpd Logging to syslog. Facility = authpriv, level = info Log_on_success flags = HOST PID Log_on_failure flags = HOST running servers = 12 ^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^ retry servers = 0 attempts = 0 service fd = 3 shutdown function = (null) Near the end: ------------- active_services = 1 available_services = 1 descriptors_free = 1015 running_servers = 261 ^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^ Logging service = enabled Shutdown service = enabled ====================================================================== Furthermore sending a "kill -ABRT" (same as SIGIOT mentioned in xinetd manual but kill on RH 7.2 doesn't recognize -IOT) and the following appears in /var/log/messages: xinetd[671]: service ftp: actual running servers = 301, known running servers = 54 xinetd[671]: Consistency check detected 1 errors I notice also quite some of the following messages in my /var/log/messages: xinetd[671]: Service ftp: server exit with 0 running servers and occasionally: xinetd[671]: Deactivating service ftp due to excessive incoming connections. Restarting in 30 seconds. Sysinfo: -------- Athlon 900 MHz, Red Hat Linux 7.2, kernel 2.4.9-7. Completely standard RH 7.2 setup (compiled vsftpd myself, but I don't think the problem is related to the ftp daemon). vsftpd is the *only* service started by xinetd. All vsftpd processes were started by the running xinetd and it has been running since boot-time. Please contact me if you need more info ! greetings, Rob van Nieuwkerk
Assigning to myself.
Can you reproduce this with the latest xinetd? I'm unable to reproduce it here, but my test environment may not be exactly the same. I believe this was a duplicate of the lost SIGCHLD problem. The packages at http://people.redhat.com/teg/xinetd/ or the source at http://www.xinetd.org/devel will give a good indication.
Yes, I can reproduce it with the latest version. I built http://people.redhat.com/teg/xinetd/xinetd-2.3.4-0.3.src.rpm I have a user who has a script that makes many download accesses to my server that triggers the bug. I asked him to rerun his script after my upgrade. Same problem. If you want I can ask him to tell what the script exactly does, so you can reproduce the problem on your system (should be very easy). Greetings, Rob van Nieuwkerk
Yes, please, I'd like the reproducing script (-0.3 is 2001-12-13, BTW)
Created attachment 40808 [details] example script that triggers xinetd bug
Hi, The (type of) script my user uses to trigger the bug is just a bunch of "ncftpget -b" commands. These start in the background so you get many connections in a short time. See the attachment for an example script. greetings, Rob van Nieuwkerk
I have seen related issues. I have per_source limits, and see FTP going well beyond this (in /var/log/secure, I see 20 or 30 incoming connections, THEN the disconnects, as a part of someone's attack on a server). I have per_source set to 3. I'm also going to submit another bug because xinetd starts doing USERID (Ident) after resetting a service following an attack. Looks like lots of bugs still lingering in xinetd.
Red Hat Linux is no longer supported by Red Hat, Inc. If you are still running Red Hat Linux, you are strongly advised to upgrade to a current Fedora Core release or Red Hat Enterprise Linux or comparable. Some information on which option may be right for you is available at http://www.redhat.com/rhel/migrate/redhatlinux/. Red Hat apologizes that these issues have not been resolved yet. We do want to make sure that no important bugs slip through the cracks. Please check if this issue is still present in a current Fedora Core release. If so, please change the product and version to match, and check the box indicating that the requested information has been provided. Note that any bug still open against Red Hat Linux on will be closed as 'CANTFIX' on September 30, 2006. Thanks again for your help.
Red Hat Linux is no longer supported by Red Hat, Inc. If you are still running Red Hat Linux, you are strongly advised to upgrade to a current Fedora Core release or Red Hat Enterprise Linux or comparable. Some information on which option may be right for you is available at http://www.redhat.com/rhel/migrate/redhatlinux/. Closing as CANTFIX.