From streiner at stargate.net Tue Mar 11 14:13:51 2003 From: streiner at stargate.net (Streiner, Justin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] max length of xinetd.conf lines? Message-ID: Is there a maximum line length in xinetd.conf, or a maximum number of entries per line? What I'm running into is I have a large number of entries set up for a service I'm running from xinetd and I have individual hosts or networks listed in the only_from line as appropriate. I tried several methods for breaking the lines up (one entry per line, one entry per line with escaped newline, etc...), but none of them seemed to work. Granted, I'm running a pretty ancient version of xinetd (2.3.3). Attempts to build a newer version of xinetd on a Solaris 7 machine that doesn't appear to have any of the IPv6 headers or libraries have been futile. This may be a temporary isuse, as the machine in question will be replaced with a newer machine running Solaris 8 pretty soon. jms From bbraun at synack.net Wed Mar 12 11:01:22 2003 From: bbraun at synack.net (Rob Braun) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] max length of xinetd.conf lines? In-Reply-To: ; from streiner@stargate.net on Tue, Mar 11, 2003 at 05:13:51PM -0500 References: Message-ID: <20030312110121.I13227@lh.synack.net> On Tue, Mar 11, 2003 at 05:13:51PM -0500, Streiner, Justin wrote: > Is there a maximum line length in xinetd.conf, or a maximum number of > entries per line? What I'm running into is I have a large number of > entries set up for a service I'm running from xinetd and I have individual > hosts or networks listed in the only_from line as appropriate. There is a maximum line length, and it can vary, depending on the system you're running on. However, 8192 characters is a good first pass at what the max line length will be. > I tried several methods for breaking the lines up (one entry per line, one > entry per line with escaped newline, etc...), but none of them seemed to > work. You should be able to use the += operator to split up lines... only_from = foo only_from += bar Rob From streiner at stargate.net Wed Mar 12 12:49:05 2003 From: streiner at stargate.net (Streiner, Justin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] max length of xinetd.conf lines? In-Reply-To: <20030312110121.I13227@lh.synack.net> References: <20030312110121.I13227@lh.synack.net> Message-ID: On Wed, 12 Mar 2003, Rob Braun wrote: > On Tue, Mar 11, 2003 at 05:13:51PM -0500, Streiner, Justin wrote: > > Is there a maximum line length in xinetd.conf, or a maximum number of > > entries per line? What I'm running into is I have a large number of > > entries set up for a service I'm running from xinetd and I have individual > > hosts or networks listed in the only_from line as appropriate. > > There is a maximum line length, and it can vary, depending on the > system you're running on. However, 8192 characters is a good first > pass at what the max line length will be. Ok, I'm well under that. The longest line is around 1600 characters long. > > I tried several methods for breaking the lines up (one entry per line, one > > entry per line with escaped newline, etc...), but none of them seemed to > > work. > > You should be able to use the += operator to split up lines... > only_from = foo > only_from += bar Ok, I'll give that a try. Thanks jms From tommcl0514 at sbcglobal.net Wed Mar 12 13:41:36 2003 From: tommcl0514 at sbcglobal.net (Tom McLoughlin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] newbie Message-ID: Here goes. I've installed RH 8. On pevious versions, I have been able to set up sendmail, imap and ipop without problems. Now I cannot setup an emial client such as Kmail because in the services section there is a statement that imap or ipop must be enabled by xinetd. I have not needed to do this before and do not know how to. In the RH8 default xinetd.conf file a line states includedir /etc/xinetd.d and I have changed the default from disable yes to disable no. I do not know what to do next. From bbraun at synack.net Wed Mar 12 13:54:06 2003 From: bbraun at synack.net (Rob Braun) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] newbie In-Reply-To: ; from tommcl0514@sbcglobal.net on Wed, Mar 12, 2003 at 01:41:36PM -0800 References: Message-ID: <20030312135406.L13227@lh.synack.net> On Wed, Mar 12, 2003 at 01:41:36PM -0800, Tom McLoughlin wrote: > Here goes. > > I've installed RH 8. On pevious versions, I have been able to set up > sendmail, imap and ipop without problems. Redhat has been shipping xinetd by default since 7.0, so versions 7.0 through 8.0 should be similar. > and I have changed the default from disable yes to disable no. > > I do not know what to do next. Have you restarted/reloaded xinetd? Redhat provides /etc/init.d/xinetd as a way to restart services. You can also send a SIGHUP to xinetd to get it to reload it's config file. Ro From btomhave at sofast.net Wed Mar 12 13:58:56 2003 From: btomhave at sofast.net (Benjamin Tomhave) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] newbie In-Reply-To: Message-ID: did you restart xinetd after doing that? > -----Original Message----- > From: xinetd-admin@xinetd.org [mailto:xinetd-admin@xinetd.org]On Behalf > Of Tom McLoughlin > Sent: Wednesday, March 12, 2003 2:42 PM > To: xinetd@xinetd.org > Subject: [xinetd] newbie > > > Here goes. > > I've installed RH 8. On pevious versions, I have been able to set up > sendmail, imap and ipop without problems. > > Now I cannot setup an emial client such as Kmail because in the services > section there is a statement that imap or ipop must be enabled by > xinetd. I > have not needed to do this before and do not know how to. > > In the RH8 default xinetd.conf file a line states > includedir /etc/xinetd.d > > and I have changed the default from disable yes to disable no. > > I do not know what to do next. > > > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > From nathanb at clinicomp.com Wed Mar 12 14:03:48 2003 From: nathanb at clinicomp.com (Nathan Bardsley) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] newbie In-Reply-To: References: Message-ID: <3E6FAEC4.50406@clinicomp.com> You need to have xinetd reload its configuration. The easy way is to use the init script that starts xinetd in the first place: sudo -c '/etc/init.d/xinetd reload', or you could do something like kill -HUP "pid of xinetd". The man page goes into some detail about the various ways of getting a new configuration active, including what happens in the case that there are more active sessions than the new limit values. Redhat has done some really good work with their init scripts providing more functionality than just stop or start. --Nathan Tom McLoughlin wrote: > Here goes. > > I've installed RH 8. On pevious versions, I have been able to set up > sendmail, imap and ipop without problems. > > Now I cannot setup an emial client such as Kmail because in the services > section there is a statement that imap or ipop must be enabled by xinetd. I > have not needed to do this before and do not know how to. > > In the RH8 default xinetd.conf file a line states > includedir /etc/xinetd.d > > and I have changed the default from disable yes to disable no. > > I do not know what to do next. > > > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > > From linux_4ever at yahoo.com Sat Mar 15 09:09:45 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd memory leaks Message-ID: <20030315170945.88755.qmail@web9605.mail.yahoo.com> Hello, I wanted to let everyone know that I've discovered several memory leaks in xinetd. I started investigating this using the valgrind memory leak testing tool. It showed a number of problems depending on how you use xinetd. The leaks are cumulative and can cause xinetd to have a large memory footprint over time. The fixes for the memory leaks are closely related to the leaked descriptors problem from last fall. The cvs copy of xinetd has all know memory leaks fixed. Since it is interrelated to the leaked descriptors, it needs careful testing by as many people as I can get. If you build & test it, look for segfaults & leaked descriptors. Periodically running: "netstat -an | grep TIME_WAIT | wc -l" will give you a good measure of fd leakage. There are tools that you can use to watch the logs for segfaults. As I said, I believe its all fixed...it just needs testing. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Web Hosting - establish your business online http://webhosting.yahoo.com From cougar at random.ee Sun Mar 16 01:25:27 2003 From: cougar at random.ee (Cougar) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] IPv6 redirect patch Message-ID: Hi, I found one logical error in redirect.c. I my case I tried to make IPv6->IPv4 proxy. In config file I had such lines: service proxy { port = 8000 bind = 2001:ad0:: socket_type = stream protocol = tcp redirect = 10.0.0.1 8000 [----] Now when I connect to 2001:ad0:: port 8000, xinetd should connect to IPv4 address. Here is what it actually does: accept(6, {sin_family=AF_INET6, sin6_port=htons(38638), inet_pton(AF_INET6, "2001:ad0:ff:6::2", &sin6_addr), sin6_flowinfo=htonl(0)}}, [28]) = 8 [----] [pid 6809] socket(PF_INET6, SOCK_STREAM, 0) = 9 [pid 6809] connect(9, {sin_family=AF_INET, sin_port=htons(8000), sin_addr=inet_addr("10.0.0.1")}}, 28) = -1 EAFNOSUPPORT (Address family not supported by protocol) In redirect.c, xinetd creates socket with the same protocol which it listened: if( SC_IPV4(scp) ) { sin_len = sizeof( struct sockaddr_in ); RedirServerFd = socket(AF_INET, SOCK_STREAM, 0); } else if( SC_IPV6(scp) ) { sin_len = sizeof( struct sockaddr_in6 ); RedirServerFd = socket(AF_INET6, SOCK_STREAM, 0); The right proto is what getaddrinfo() gives for destination address. So, the right code should be: if( serveraddr.sa_in.sin_family == AF_INET ) { sin_len = sizeof( struct sockaddr_in ); RedirServerFd = socket(AF_INET, SOCK_STREAM, 0); } else if( serveraddr.sa_in.sin_family == AF_INET6 ) { sin_len = sizeof( struct sockaddr_in6 ); RedirServerFd = socket(AF_INET6, SOCK_STREAM, 0); This way it can redirect IPv6 sockets to both IPv4 and IPv6 addresses. t works also with IPv4 sockets without any problem. I tested all four situations. I attached path to this mail. There are two places in redirect.c which should be changed. --- Cougar -------------- next part -------------- --- xinetd-2.3.10.orig/xinetd/redirect.c Mon Sep 9 22:40:52 2002 +++ xinetd-2.3.10/xinetd/redirect.c Sun Mar 16 10:57:35 2003 @@ -72,10 +72,11 @@ if( scp->sc_protocol.value == IPPROTO_TCP ) { char *foo = NULL; + memcpy(&serveraddr, scp->sc_redir_addr, sizeof(serveraddr)); - if( SC_IPV4(scp) ) { + if( serveraddr.sa_in.sin_family == AF_INET ) { sin_len = sizeof( struct sockaddr_in ); RedirServerFd = socket(AF_INET, SOCK_STREAM, 0); - } else if( SC_IPV6(scp) ) { + } else if( serveraddr.sa_in.sin_family == AF_INET6 ) { sin_len = sizeof( struct sockaddr_in6 ); RedirServerFd = socket(AF_INET6, SOCK_STREAM, 0); } else { @@ -95,10 +96,9 @@ msg(LOG_ERR, func, "setsockopt SO_KEEPALIVE RedirServerFd failed: %m"); - memcpy(&serveraddr, scp->sc_redir_addr, sizeof(serveraddr)); - if( SC_IPV4(scp) ) + if( serveraddr.sa_in.sin_family == AF_INET ) serveraddr.sa_in.sin_port = htons(serveraddr.sa_in.sin_port); - if( SC_IPV6(scp) ) + if( serveraddr.sa_in.sin_family == AF_INET6 ) serveraddr.sa_in6.sin6_port = htons(serveraddr.sa_in6.sin6_port); if( connect(RedirServerFd, &serveraddr.sa, sin_len) < 0 ) From tolivier at rightvision.com Tue Mar 18 03:22:10 2003 From: tolivier at rightvision.com (Tchilinguirian Olivier) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] Process defunct... Message-ID: <002d01c2ed40$b31e4330$0500000a@otb> Hi I have some troubles with xinetd on my server (redhat 7.1 default). I'm using a webmail application which does some imap requests... Sometimes (often), the webmail app. doesn't answer because of imapd defunct process owned by xinted. If I send a sig-child to xinetd, everything goes well and the defunct process die, and IMAP is once again reachable. So I take a look to the xinetd sources in order to find the guilty routine ;-) Finaly I found that the problem comes from "child_exit" routine in "child.c". The "waitpid" is blocked and imap service remains unreachable (interminable). I modified the waitpid call by adding the "WUNTRACED" option... pid = waitpid( -1, &status, WNOHANG|WUNTRACED ); This change seems to fix the problem... I tested the IMAP service by making a lot of request and I doesn't have any other defunct imap process. Can you take a look and tell me if these changes won't bring up new bugs ;-) Thanks, Olivier -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xinetd.org/pipermail/xinetd/attachments/20030318/83dfed14/attachment.html From linux_4ever at yahoo.com Tue Mar 18 04:56:06 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] Process defunct... In-Reply-To: <002d01c2ed40$b31e4330$0500000a@otb> Message-ID: <20030318125606.63438.qmail@web9608.mail.yahoo.com> >I modified the waitpid call by adding the >WUNTRACED" option... This is for job control. Why is your imap program stopping instead of exiting? I suppose it solves your problem, but it shows there's a basic design flaw in the imap program. i(x)inetd started programs are supposed to run to completion and exit. If you need to add WUNTRACED, your program has not exited. Since your application is not answering, it also sounds like you are running this as a wait service, which is mostly intended for udp sockets. If wait = no, then xinetd should continue listening and handling requests on whatever port you set it up on. Offhand, I'd say that your problem is either configuration of the service or the imap program is not a well behaved inetd style program. If you don't mind, which imap program is it? -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From linux_4ever at yahoo.com Tue Mar 18 05:00:32 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] IPv6 redirect patch In-Reply-To: Message-ID: <20030318130032.73231.qmail@web9604.mail.yahoo.com> >I found one logical error in redirect.c. I saw this last summer when I was working on redirect & I think I mentioned something like this to Rob, I've since deleted those e-mails & I don't remember what the reason was as to why it wasn't changed. Rob, do you have any objections with this patch? It looks OK to me. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From Patrick.Dewilde at hp.com Tue Mar 18 09:21:21 2003 From: Patrick.Dewilde at hp.com (Dewilde, Patrick) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) Message-ID: <2B90F32C46F7D043B05AF7F700C4B1BB01B29528@broexc02.emea.cpqcorp.net> Steve, We were planning to use xinetd to redirect connections on a RedHat system kernel 2.4, which needs to sustain 3000 concurrent redirected sessions (6000 established connections inbound/outbound). Do you think the memory leaks you discovered will impact the stability under such redirect load? Thanks, patrick -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.xinetd.org/pipermail/xinetd/attachments/20030318/bf7c0f64/attachment.html From bbraun at synack.net Tue Mar 18 09:49:06 2003 From: bbraun at synack.net (Rob Braun) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) In-Reply-To: <2B90F32C46F7D043B05AF7F700C4B1BB01B29528@broexc02.emea.cpqcorp.net>; from Patrick.Dewilde@hp.com on Tue, Mar 18, 2003 at 06:21:21PM +0100 References: <2B90F32C46F7D043B05AF7F700C4B1BB01B29528@broexc02.emea.cpqcorp.net> Message-ID: <20030318094905.J9799@lh.synack.net> On Tue, Mar 18, 2003 at 06:21:21PM +0100, Dewilde, Patrick wrote: > Steve, > > We were planning to use xinetd to redirect connections on a RedHat system kernel 2.4, which needs to sustain 3000 concurrent redirected sessions (6000 established connections inbound/outbound). I have to ask why you desire to use xinetd for such redirections. Linux systems have kernel level redirects which are far more efficient than bouncing through a user level process. I have always found it rather silly to use xinetd's redirect feature on an OS that supports redirection natively, and am personally curious what about the xinetd redirection makes it more desireable in your case? Rob From linux_4ever at yahoo.com Tue Mar 18 10:02:12 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) In-Reply-To: <2B90F32C46F7D043B05AF7F700C4B1BB01B29528@broexc02.emea.cpqcorp.net> Message-ID: <20030318180212.99078.qmail@web9606.mail.yahoo.com> Hello, >Do you think the memory leaks you discovered will >impact the stability under such redirect load? This commit here: http://www.xinetd.org/pipermail/cvs-xinetd/2003-March/000058.html is the main one fixing the memory leaks. Some of the commit was fixing wrapping text, but the piece to study is around chunk 551. It seems that when a connection is rejected, internal, or a wait service, the connection's memory for ancillary data was not freed. The File Descriptor was closed, though. Note that this is cumulative. Each rejected connection leaks 144 bytes. There were a number of updates to sconf.c, too, that fixed memory leaks whenever a hard reconfig was issued via SIGHUP. These are not as problematic. So, if you have a situation where people could be rejected connections (only_from, tcp_wrappers, no_access, access_times, cps, sensors, etc)...yes it could affect you. If you have a nice environment with no one playing games or no restrictions on access, then you are probably OK. I was motivated to find this after a ftp worm hit my server & did 5000+ connections in 1 second which killed xinetd. I only allow ftp to & from a couple domains. My server is now running the cvs version with no ill effects. So far, I haven't heard good or bad feedback about the call for testing of the cvs version, but it could impact users. The sooner people confirm its OK, the better. I personally would like to see a new xinetd release soon, but its Rob's call as to when that is. Hope this helps... -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From cougar at random.ee Tue Mar 18 10:09:47 2003 From: cougar at random.ee (Cougar) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) In-Reply-To: <20030318180212.99078.qmail@web9606.mail.yahoo.com> Message-ID: On Tue, 18 Mar 2003, Steve G wrote: > This commit here: > http://www.xinetd.org/pipermail/cvs-xinetd/2003-March/000058.html > is the main one fixing the memory leaks. Some of the commit [----] - memcpy(sp->svc_last_dgram_addr, sinp, sizeof(struct sockaddr_in)); + memcpy(sp->svc_last_dgram_addr, sinp,sizeof(struct sockaddr_in)); - memcpy(sp->svc_last_dgram_addr, sinp, sizeof(struct sockaddr_in6)); + memcpy(sp->svc_last_dgram_addr,sinp,sizeof(struct sockaddr_in6)); Why are such lines there ;-) --- Cougar From linux_4ever at yahoo.com Tue Mar 18 10:15:40 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) In-Reply-To: Message-ID: <20030318181540.20555.qmail@web9608.mail.yahoo.com> Hello, Everything except chunck 551 is fixing wrapping text. This has been there for years & I wanted to get rid of it since I was going to make a commit. Wrapping text messes up readability & printing of source code. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From cougar at random.ee Tue Mar 18 10:36:32 2003 From: cougar at random.ee (Cougar) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] (no subject) In-Reply-To: <20030318181540.20555.qmail@web9608.mail.yahoo.com> Message-ID: On Tue, 18 Mar 2003, Steve G wrote: > Everything except chunck 551 is fixing wrapping text. This > has been there for years & I wanted to get rid of it since > I was going to make a commit. Wrapping text messes up > readability & printing of source code. Yes, I got the point. It was just a strange where there was space after one comma in first change and no spaces in second one. For me it doesn't make a code more clear.. ;-) (IMHO Linux kernel has one of the best coding style) --- Cougar From murygin at aitoc.com Tue Mar 18 11:44:47 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam Message-ID: <20030318214447.59726160.murygin@aitoc.com> Hi, I'm trying to run fam through xinetd-2.3.10 or through today CVS, but it doesn't work. It's just write following info into log. Mar 18 10:32:51 lcd-semen xinetd[14068]: Deactivating service sgi_fam due to excessive incoming connections. Restarting in 10 seconds. ] Mar 18 10:33:02 lcd-semen xinetd[14068]: Activating service sgi_fam But if i try run it through xinetd-2.3.7, it's worked :( From linux_4ever at yahoo.com Tue Mar 18 12:14:24 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318214447.59726160.murygin@aitoc.com> Message-ID: <20030318201424.51691.qmail@web9604.mail.yahoo.com> Hello, What's your configuration and how are you testing it? -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From murygin at aitoc.com Tue Mar 18 12:18:23 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318201424.51691.qmail@web9604.mail.yahoo.com> References: <20030318214447.59726160.murygin@aitoc.com> <20030318201424.51691.qmail@web9604.mail.yahoo.com> Message-ID: <20030318221823.37db21b0.murygin@aitoc.com> On Tue, 18 Mar 2003 12:14:24 -0800 (PST) Steve G wrote: > Hello, > > What's your configuration and how are you testing it? I test it with fileschanged (http://fileschanged.sourceforge.net/). xinetd configured with libwrap and loadavg fam is 2.6.9. what other info do you need? Thanks. > > -Steve Grubb > > __________________________________________________ > Do you Yahoo!? > Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! > http://platinum.yahoo.com > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > From linux_4ever at yahoo.com Tue Mar 18 12:41:46 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318221823.37db21b0.murygin@aitoc.com> Message-ID: <20030318204146.93315.qmail@web9602.mail.yahoo.com> >what other info do you need? The services configuration settings. It is either in /etc/xinetd.d or /etc/xinetd.conf. Just the settings for that service. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From murygin at aitoc.com Tue Mar 18 12:41:57 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318204146.93315.qmail@web9602.mail.yahoo.com> References: <20030318221823.37db21b0.murygin@aitoc.com> <20030318204146.93315.qmail@web9602.mail.yahoo.com> Message-ID: <20030318224157.760e4512.murygin@aitoc.com> On Tue, 18 Mar 2003 12:41:46 -0800 (PST) Steve G wrote: > >what other info do you need? > > The services configuration settings. It is either in > /etc/xinetd.d or /etc/xinetd.conf. Just the settings for > that service. there two config one from my distro service sgi_fam { disable = no socket_type = stream protocol = tcp type = RPC rpc_version = 2 bind = 127.0.0.1 wait = yes user = root server = /usr/sbin/fam } but when it didn't work, i try with another one service sgi_fam { disable = no flags = NOLIBWRAP type = RPC UNLISTED socket_type = stream user = root group = nobody server = /usr/sbin/fam wait = yes protocol = tcp rpc_version = 2 rpc_number = 391002 } but it didn't work as well :( > > -Steve Grubb > > > __________________________________________________ > Do you Yahoo!? > Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! > http://platinum.yahoo.com > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > From linux_4ever at yahoo.com Tue Mar 18 13:29:30 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318224157.760e4512.murygin@aitoc.com> Message-ID: <20030318212930.60942.qmail@web9607.mail.yahoo.com> The second configuration is correct. You have portmapper starting before xinetd & rpcinfo -p shows its alive? rpcinfo -t localhost sgi_fam does anything for you? Does rpcinfo make same messages in the logs? -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From murygin at aitoc.com Tue Mar 18 13:37:39 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318212930.60942.qmail@web9607.mail.yahoo.com> References: <20030318224157.760e4512.murygin@aitoc.com> <20030318212930.60942.qmail@web9607.mail.yahoo.com> Message-ID: <20030318233739.61516c92.murygin@aitoc.com> On Tue, 18 Mar 2003 13:29:30 -0800 (PST) Steve G wrote: > The second configuration is correct. You have portmapper > starting before xinetd & rpcinfo -p shows its alive? Yes > rpcinfo -t localhost sgi_fam does anything for you? Does > rpcinfo make same messages in the logs? No, when i start fam manually, it's works ok sorry, bit it's too late for me, i will be tommorow :) thanks for your answers > > -Steve Grubb > > __________________________________________________ > Do you Yahoo!? > Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! > http://platinum.yahoo.com > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > From linux_4ever at yahoo.com Tue Mar 18 15:22:03 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318233739.61516c92.murygin@aitoc.com> Message-ID: <20030318232203.20037.qmail@web9602.mail.yahoo.com> Hello, What I'm seeing is if xinetd rejects the connection, it will loop forever in the current cvs version. I also see a leak of memory when rpc services are run. I will submit a patch later that solves this last memory leak. I also noticed that portmapper is fragile and sometimes needs restarting, too. Stay with the second config, make sure portmap is in hosts.allow file and see if it all boots up OK tomorrow. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From murygin at aitoc.com Wed Mar 19 03:35:10 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318212930.60942.qmail@web9607.mail.yahoo.com> References: <20030318224157.760e4512.murygin@aitoc.com> <20030318212930.60942.qmail@web9607.mail.yahoo.com> Message-ID: <20030319133510.2d92281c.murygin@aitoc.com> On Tue, 18 Mar 2003 13:29:30 -0800 (PST) Steve G wrote: > The second configuration is correct. You have portmapper > starting before xinetd & rpcinfo -p shows its alive? > rpcinfo -t localhost sgi_fam does anything for you? Does > rpcinfo make same messages in the logs? there two attached logs first (xinetd_notwork.log) xinetd-2.3.10 second (xinetd_work.log) xinetd-2.3.7 I hope it will help you :) > -Steve Grubb > > __________________________________________________ > Do you Yahoo!? > Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! > http://platinum.yahoo.com > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > -------------- next part -------------- A non-text attachment was scrubbed... Name: xinetd_notwork.log Type: application/octet-stream Size: 9349 bytes Desc: not available Url : http://www.xinetd.org/pipermail/xinetd/attachments/20030319/93bb4086/xinetd_notwork.obj -------------- next part -------------- A non-text attachment was scrubbed... Name: xinetd_work.log Type: application/octet-stream Size: 6723 bytes Desc: not available Url : http://www.xinetd.org/pipermail/xinetd/attachments/20030319/93bb4086/xinetd_work.obj From tolivier at rightvision.com Wed Mar 19 06:29:24 2003 From: tolivier at rightvision.com (Tchilinguirian Olivier) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] Process defunct... References: <20030318125606.63438.qmail@web9608.mail.yahoo.com> Message-ID: <002a01c2ee24$073d81e0$0500000a@otb> Thanks for your answer, I'll try to investigate on imap side... Effectively, if I need to add WUNTRACED it's because IMAP process doesn't exit properly... But even if my service is configured with "wait = no", when this error occures, the imap service is completely KO. For info: my imap daemon comes from the redhat 7.1 rpms => imap-2000c-10 I'll take a look to redhat bugtracking about this imap issue... Olivier. From linux_4ever at yahoo.com Wed Mar 19 07:19:13 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] Process defunct... In-Reply-To: <002a01c2ee24$073d81e0$0500000a@otb> Message-ID: <20030319151913.98061.qmail@web9608.mail.yahoo.com> >Effectively, if I need to add WUNTRACED it's because IMAP >process doesn't exit properly... Actually, this means imap received a SIGSTOP. For this service, you should use socket_type = stream, wait = no, user = root. I also have nice = 10, but you may not need that. Have you tried connecting to imap without your web application? SIGPIPE in the web application may cause problems. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From tolivier at rightvision.com Wed Mar 19 08:19:39 2003 From: tolivier at rightvision.com (Tchilinguirian Olivier) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] Process defunct... References: <20030319151913.98061.qmail@web9608.mail.yahoo.com> Message-ID: <002101c2ee33$7520c5a0$0500000a@otb> You are probably right about the SIGPIPE ... Concerning the service config., I'm using exactly the same... I tried to connect to imap with telnet ("telnet localhost 143") when service hanged and I got a blank prompt...bizar ;-) Olivier. From Listen at e-admin.de Wed Mar 19 14:04:26 2003 From: Listen at e-admin.de (Andreas Gietl) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] feature request: inclusion of files for only_from and no_access Message-ID: <200303192304.26805.Listen@e-admin.de> Hi, I'd like to see xinetd to be able to not only use "static" host and network definitions for only_from and no_access, but use dynamic a dynamik host-list by inclusion of an external file. This would make us capable of using the same file we use for pop before smtp to restrict access to the sslwrapped instances of the mail server. And of course many other useful things may be done with this feature. This would of course slow down performance a little bit, but everyone is free to use or not use this feature. Do you think this is a good idea, or a bad one? Would be cool if i would see this feature in the announcement of one of the next releases. Andreas -- e-admin internet gmbh Andreas Gietl tel +49 941 3810884 Ludwig-Thoma-Strasse 35 fax +49 89 244329104 93051 Regensburg mobil +49 171 6070008 PGP/GPG-Key unter http://www.e-admin.de/gpg.html From murygin at aitoc.com Fri Mar 21 11:33:50 2003 From: murygin at aitoc.com (Alex Murygin) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd & fam In-Reply-To: <20030318212930.60942.qmail@web9607.mail.yahoo.com> References: <20030318224157.760e4512.murygin@aitoc.com> <20030318212930.60942.qmail@web9607.mail.yahoo.com> Message-ID: <20030321213350.2721b80c.murygin@aitoc.com> On Tue, 18 Mar 2003 13:29:30 -0800 (PST) Steve G wrote: I'm trying it from today CVS, its worked :)) Thanks very much > The second configuration is correct. You have portmapper > starting before xinetd & rpcinfo -p shows its alive? > rpcinfo -t localhost sgi_fam does anything for you? Does > rpcinfo make same messages in the logs? > > -Steve Grubb > > __________________________________________________ > Do you Yahoo!? > Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! > http://platinum.yahoo.com > _______________________________________________ > xinetd mailing list > xinetd@xinetd.org > http://www.xinetd.org/mailman/listinfo/xinetd > From juergen.edner at telejeck.de Sat Mar 22 04:15:04 2003 From: juergen.edner at telejeck.de (Juergen Edner) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found Message-ID: <000501c2f06c$ac17f2a0$0201a8c0@privatnet.lan> Hello, I installed the talk daemon on my machine an got the follwing error when I use the service. Right after that the talkd cannot be used any more. xinetd v.2.3.10 and cvs20030322 START: talk file descriptor of service talk has been closed select reported EBADF but no bad file descriptors were found I googled for this error message but couldn't find a solution for this problem. They always pointed out that it seams to be an xinetd problem. If anyone can give me a hint how to solve this problem it will be apreciated. Regards Juergen -- Mail: juergen.edner@telejeck.de PGP Key available --- From linux_4ever at yahoo.com Sat Mar 22 05:34:54 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:50 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: <000501c2f06c$ac17f2a0$0201a8c0@privatnet.lan> Message-ID: <20030322133454.71636.qmail@web9608.mail.yahoo.com> Hello, I've seen this many times as I'ver been troublesooting the memory leaks. It usually means a file descriptor was closed that shouldn't have been closed. Send the talk service config & I'll check it here. Its easy to fix once found. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From juergen.edner at telejeck.de Sun Mar 23 02:55:31 2003 From: juergen.edner at telejeck.de (Juergen Edner) Date: Thu Oct 6 10:34:51 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: <20030322200001.5882.9983.Mailman@lh.synack.net> Message-ID: <000101c2f12a$b99f23d0$0201a8c0@privatnet.lan> Hello Steve, > I've seen this many times as I'ver been troublesooting the > memory leaks. It usually means a file descriptor was closed > that shouldn't have been closed. > > Send the talk service config & I'll check it here. Its easy > to fix once found. I will try to send you as much information as I can which may helps you solving the problem :-) Environment: Linux Kernel 2.2.19 netkit-ntalk-0.17 xinetd v2.3.10 and v20030122 tried talk service config: service talk { socket_type = dgram protocol = udp wait = no user = root server = /usr/sbin/talkd log_on_success += USERID log_on_failure += USERID disable = no } Scenario: I tried to contact a WinTalk v1.27 32-bit client by using ytalk v3.1.1 from my linux box. The first communication generates the following error message: xinetd[23757]: START: talk xinetd[23757]: file descriptor of service talk has been closed xinetd[23757]: select reported EBADF but no bad file descriptors were found After an indefined number of successful connections and some additional time without a connection attempt the talk daemon refuses any further connection attempts. Regards J?rgen -- Mail: juergen.edner@telejeck.de PGP Key available --- From linux_4ever at yahoo.com Sun Mar 23 09:48:13 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:51 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: <000101c2f12a$b99f23d0$0201a8c0@privatnet.lan> Message-ID: <20030323174813.97417.qmail@web9603.mail.yahoo.com> > wait = no udp sockets are usually wait = yes. Try changing it and see if that fixes it. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com From juergen.edner at telejeck.de Mon Mar 24 09:52:12 2003 From: juergen.edner at telejeck.de (Juergen Edner) Date: Thu Oct 6 10:34:51 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: <20030323200001.17845.48829.Mailman@lh.synack.net> Message-ID: <000001c2f22e$1a423840$0201a8c0@privatnet.lan> Hello Steve, > > wait = no > > udp sockets are usually wait = yes. Try changing it and see > if that fixes it. two thumbs up for this hint. The problem seems to be solved now. I'm not sure why I haven't tried to modify the parameter, maybe because the error message pointed in the wrong direction. Again, thank you for your fast help! cu J?rgen -- Mail: juergen.edner@telejeck.de PGP Key available --- From ma+xi at dt.e-technik.uni-dortmund.de Mon Mar 24 11:38:10 2003 From: ma+xi at dt.e-technik.uni-dortmund.de (Matthias Andree) Date: Thu Oct 6 10:34:51 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: <000001c2f22e$1a423840$0201a8c0@privatnet.lan> ("Juergen Edner"'s message of "Mon, 24 Mar 2003 18:52:12 +0100") References: <000001c2f22e$1a423840$0201a8c0@privatnet.lan> Message-ID: "Juergen Edner" writes: > two thumbs up for this hint. The problem seems to be solved now. > I'm not sure why I haven't tried to modify the parameter, maybe > because the error message pointed in the wrong direction. Steve, how about this patch (against current CVS): Index: ./xinetd/xinetd.conf.man =================================================================== RCS file: /cvs/xinetd/xinetd/xinetd/xinetd.conf.man,v retrieving revision 1.5 diff -u -r1.5 xinetd.conf.man --- ./xinetd/xinetd.conf.man 23 Mar 2003 00:35:11 -0000 1.5 +++ ./xinetd/xinetd.conf.man 24 Mar 2003 19:35:35 -0000 @@ -181,6 +181,8 @@ handling requests for the service until the server dies. If the attribute value is \fIno\fP, the service is multi-threaded and \fBxinetd\fP will keep handling new service requests. +Note that sockets of type \fIdgram\fP usually require this attribute set +to \fIyes\fP. .TP .B user determines the uid for the server process. The user name must exist in -- Matthias Andree From linux_4ever at yahoo.com Tue Mar 25 05:35:17 2003 From: linux_4ever at yahoo.com (Steve G) Date: Thu Oct 6 10:34:51 2005 Subject: [xinetd] xinetd - select reported EBADF but no bad file descriptors were found In-Reply-To: Message-ID: <20030325133517.59910.qmail@web9608.mail.yahoo.com> >Steve, how about this patch (against current CVS I think you are right that this needs addressing. I will submit a more detailed explanation of what wait really means later today. In short, it has 2 effects. People should know what both of these are in order to make better decisions. -Steve Grubb __________________________________________________ Do you Yahoo!? Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop! http://platinum.yahoo.com