inetd seems to lock-up

inetd seems to lock-up

Post by olav woelfelschneid » Mon, 03 Oct 1994 04:27:36



While trying to connect a linux box to the internet i got a strange
behaviour. The machine first accepts telnet or any other inetd
initiated connection, while a few minutes later doing the same thing
will hang. Thus, trying to telnet into the box yields the
connected to message, but there is no prompt.
This happens with every connection which is handled by inetd,
like ftp, finger, etc...
After a long delay of maybe several hours (maybe only a few minutes)
the machine again accepts connections, but after a short time it hangs
again.

Weirdly, a lpmud driver running permanently listening to port 7680 runs
perfectly, thus i guess the fault has something to do with inetd.

I know about the order hosts, bind problem in /etc/resolv.conf, but this
should not matter, since there is a working nameserver on the local
network.

Anyone had that problem, too and solved it?

The machine:
486DX33/16MB, seagate 120MB ide drive, wd80*0 network card running
linux 1.1.51 with the tcpip package taken from slackware 2.0

Thanx,
  Olav
--
/======================================+======================================\
| Olav "Mac" Woelfelschneider          | I refuse to grow up,                 |

\======================================+======================================/

 
 
 

inetd seems to lock-up

Post by Steve Kneiz » Mon, 03 Oct 1994 03:42:13


: While trying to connect a linux box to the internet i got a strange
: behaviour. The machine first accepts telnet or any other inetd
: initiated connection, while a few minutes later doing the same thing
: will hang. Thus, trying to telnet into the box yields the
: connected to message, but there is no prompt.
: This happens with every connection which is handled by inetd,
: like ftp, finger, etc...
: After a long delay of maybe several hours (maybe only a few minutes)
: the machine again accepts connections, but after a short time it hangs
: again.

: Weirdly, a lpmud driver running permanently listening to port 7680 runs
: perfectly, thus i guess the fault has something to do with inetd.

: I know about the order hosts, bind problem in /etc/resolv.conf, but this
: should not matter, since there is a working nameserver on the local
: network.

: Anyone had that problem, too and solved it?

: The machine:
: 486DX33/16MB, seagate 120MB ide drive, wd80*0 network card running
: linux 1.1.51 with the tcpip package taken from slackware 2.0

: Thanx,
:   Olav

Heya Olav,

Another case of it!  I have been waiting for several weeks now for an
answer on this one, but nobody has come up with anything.  At first
I thought it was something I did wrong, but it does seem to be a
problem with the code.  Where exactly, other than with inetd, I do not know.  

I just put one of my boxes to 1.0.8 to clear this up, and the
Pentium EISA/PCI with 2 Gig SCSI box is sitting in a corner waiting
for this fix.  I have seen this bug as far back as 1.1.48, but I
cannot determine how far back it goes ... would try to find out if
a developer asked.

Others have emailed me about it too, no solutions ... be sure
to post if you find one!

Good Luck,

Steve...

 
 
 

inetd seems to lock-up

Post by Lee J. Silverm » Tue, 04 Oct 1994 07:11:56


    Just so you know, I have a Linux box running Slackware 1.2,
Binutils 1.5, and kernel version 1.1.37, and my Inetd hangs every once
in a while as well.  When this happens, I can connection to my web,
gopher, ftp, and smail daemons, because they all run independently of
inetd, but I cannot log in or rlogin.  The problem usually seems to
correct itself after an hour or two.  If you have received *any* word
of a fix I'd love to hear about it... my console often isn't
accessible so when inetd is down there's no way to get into the
machine...

Thanks!
--
Lee Silverman, Brown class of '94, Brown GeoPhysics ScM '95


"Nonsense - you only say it's impossible because nobody's ever done it."

 
 
 

inetd seems to lock-up

Post by Jarkko Ait » Tue, 04 Oct 1994 12:09:08




Quote:

>    Just so you know, I have a Linux box running Slackware 1.2,
>Binutils 1.5, and kernel version 1.1.37, and my Inetd hangs every once
>in a while as well.  When this happens, I can connection to my web,
>gopher, ftp, and smail daemons, because they all run independently of
>inetd, but I cannot log in or rlogin.  The problem usually seems to
>correct itself after an hour or two.

Hi, we had one few months old slackware installation here which did just that
what you described, only not so often. Now that machine is running with inetd
which i compiled from some net package long ago (can't remember which) and no
problems this far. (few days now)

-AJake

 
 
 

inetd seems to lock-up

Post by Bradley M. Ku » Wed, 05 Oct 1994 09:52:16





>: While trying to connect a linux box to the internet i got a strange
>: behaviour. The machine first accepts telnet or any other inetd
>: initiated connection, while a few minutes later doing the same thing
>: will hang. Thus, trying to telnet into the box yields the
>: connected to message, but there is no prompt.
>: This happens with every connection which is handled by inetd,
>: like ftp, finger, etc...

>Another case of it!  I have been waiting for several weeks now for an
>answer on this one, but nobody has come up with anything.  At first
>I thought it was something I did wrong, but it does seem to be a
>problem with the code.  Where exactly, other than with inetd, I do not know.  

>I just put one of my boxes to 1.0.8 to clear this up, and the
>Pentium EISA/PCI with 2 Gig SCSI box is sitting in a corner waiting
>for this fix.  I have seen this bug as far back as 1.1.48, but I
>cannot determine how far back it goes ... would try to find out if
>a developer asked.

>Others have emailed me about it too, no solutions ... be sure
>to post if you find one!

Alright.  I have it TOO.  I am using 1.0.9 from Slackware.  Does 1.0.8 solve
the problem (or not have it, rather)?  To think one version before would save
me these headaches!

I also have another problem.  inetd on a machine that has drives nfs mounted
from it, inetd seems to spawn extra rpc.portmaps, one after another until the
original one is killed.  is this a related problem, or something else?

--
 -- Bradley M. Kuhn

 
 
 

inetd seems to lock-up

Post by Bart Kin » Thu, 06 Oct 1994 15:59:59



>Path: otago.ac.nz!canterbury.ac.nz!waikato!ames!hookup!swrinde!howland.reston.ans.net!news.sprintlink.net!pipex!sunic!trane.uninett.no!eunet.no!nuug!EU.net!news.eunet.fi!not-for-mail

>Newsgroups: comp.os.linux.admin
>Subject: Re: inetd seems to lock-up
>Date: 3 Oct 1994 05:09:08 +0200
>Organization: Muncca Ry
>Lines: 16



>NNTP-Posting-Host: snafu.muncca.fi


>>    Just so you know, I have a Linux box running Slackware 1.2,
>>Binutils 1.5, and kernel version 1.1.37, and my Inetd hangs every once
>>in a while as well.  When this happens, I can connection to my web,
>>gopher, ftp, and smail daemons, because they all run independently of
>>inetd, but I cannot log in or rlogin.  The problem usually seems to
>>correct itself after an hour or two.
>Hi, we had one few months old slackware installation here which did just that
>what you described, only not so often. Now that machine is running with inetd
>which i compiled from some net package long ago (can't remember which) and no
>problems this far. (few days now)
>-AJake

Yes, we have the problem too. I am running a 4 line SLIP dial-in Internet
Service. Original Slackware 1.2 installation, now with kernel 1.1.50.
The problem is only with services accessed through *inetd*, and in our case
especially seems to occur when more than about 10 sockets are open, mostly
with incoming SMTP traffic.  Any attempt to get into the server with SMTP,
POP3 and TELNET (others maybe too) will simply 'hang' for a very long time,
after which it suddenly releases itself.  The problem is, that many users
simply hang-up in this situation, and re-dial in again. This causes a flood of
open sockets, which stay open for very long times, making the system
inaccesible.  Because most of our trafic is SMTP, I am considering running the
SMTPd daemon sepparate from inetd. Maybe this will overcome the situation in
our case. I hope that somebody with knowlage can look intoi this situation!

Greetings,

------------------------------------------------------------------------------------
Bart Kindt (ZL4FOX) System Operator, Efficient Software NZ LTD, Dunedin, New Zealand
------------------------------------------------------------------------------------

 
 
 

inetd seems to lock-up

Post by Matt Midb » Thu, 06 Oct 1994 02:47:54



[stuff about inetd not running processes anymore, but net connects on
non-inetd ports still work]

We have the exact same problem here. A 1.1.51 kernel (but it did it
with earlier kernels). The sum on my inetd is 63545    18. Is that
same as everyone elses? Two IDE hard drives, 500M and 330M, with a
486DX2-66 processor. They usually seem to come along with a whole
bunch of named: connection refused or tcpd: application returned 0xe
errors. (Or errors to that effect)

Any solutions to this are desperately wanted!

Matt

 
 
 

inetd seems to lock-up

Post by David La Cro » Sat, 08 Oct 1994 00:46:36



: Alright.  I have it TOO.  I am using 1.0.9 from Slackware.  Does 1.0.8 solve
: the problem (or not have it, rather)?  To think one version before would save
: me these headaches!

: I also have another problem.  inetd on a machine that has drives nfs mounted
: from it, inetd seems to spawn extra rpc.portmaps, one after another until the
: original one is killed.  is this a related problem, or something else?

The fix to this is either a) compiling with the I_am_a_broken_BSD_weenie
option or patching the libc to 2.5.26a or later.
(the rpc.portmap bug)  Also I recommend upgrading to the latest rpc.portmapper
because it fixed a couple of other potential bugs (specifically security
concerns)

: --
:  -- Bradley M. Kuhn

--


 
 
 

inetd seems to lock-up

Post by Ms. Stephanie Gilg » Sat, 08 Oct 1994 22:16:02







>[stuff about inetd not running processes anymore, but net connects on
>non-inetd ports still work]

>We have the exact same problem here. A 1.1.51 kernel (but it did it
>with earlier kernels). The sum on my inetd is 63545    18. Is that
>same as everyone elses? Two IDE hard drives, 500M and 330M, with a

Same sum here on a 1.1.33 kernel, and same problem on two machines running it.
Since I know it's there, killing/restarting inetd is a bit of a pain only,
due to frequency. I also think after time passes it will come alive again,
based on logfile data. Would be nice to be fixed tho...

                                Steph

--

Wizvax Communications - Public Access Internet - HTTP://wizvax.com/index.html
1508 Tibbits Ave. Troy, NY 12180 V:(518)271-6005 F:(518)271-6289
"There is no sound, quite like the purring of a thousand cats." - Me

 
 
 

inetd seems to lock-up

Post by David F » Sun, 09 Oct 1994 03:23:48



] "There is no sound, quite like the purring of a thousand cats." - Me

Are you sure you want a comma in this sentence?  It reads
a little strange that way.
--
David Fox                                               xoF divaD
NYU Media Research Lab                     baL hcraeseR aideM UYN