Re: MSIE automatic proxy config

2000-08-27 Thread Gilad Ben-Yossef

Shachar Shemesh wrote:
 
 You will find that your solution forwards ALL outbound packets to the proxy
 machine. Not just those aimed at port 80. You are then left with my original
 problem - I don't want to penalise the entire office traffic with an extra hop
 (actually - extra two hops and a routing loop in your solution), just because
 I want to implement a transperant proxy. A much simpler solution for me is to
 block all communication to port 80 outbound, and force everyone to manually
 configure the proxy or they don't get web access.
 

And once again I must say: "Don't think so 3rd layer, JeanLuke".

I was about to explain how to build a 2d level (OSI) bridiging proxy but
someone already did:
http://perso.wanadoo.fr/magpie/EtherDivert.html

No extra hop, no need for another subnet, batteries not included...

I do suggest however, you use the new bridge patch ported from
2.4.0-testx that can be found at http://www.openrock.net/bridge and not
the original 2.2.x bridiging code.

Gilad.
-- 
Gilad Ben-Yossef [EMAIL PROTECTED] 
http://kagoor.com | +972(9)9565333 x230 | +972(54)756701
"I've been seduced by the chocolate side of the force."

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: MSIE automatic proxy config

2000-08-27 Thread Shachar Shemesh



Gilad Ben-Yossef wrote:

 And once again I must say: "Don't think so 3rd layer, JeanLuke".

I am not, number 1.


 I was about to explain how to build a 2d level (OSI) bridiging proxy but
 someone already did:
 http://perso.wanadoo.fr/magpie/EtherDivert.html

 No extra hop, no need for another subnet, batteries not included...

Oh, but you do add an extra hop. The fact that no IP protocol is aware of that does
not change the fact that you now require all your traffic to be directed through
your box. The box still acts as a router (actually, a bridge, but same difference),
and the performance penalties are still being payed (though I have to admit that
it's probably less of a penalty).


 I do suggest however, you use the new bridge patch ported from
 2.4.0-testx that can be found at http://www.openrock.net/bridge and not
 the original 2.2.x bridiging code.

 Gilad.

Actually, I'll stick with forcing everyone to move to an explicit proxy by means of
filtering.

I do have access to the router.

Shachar



=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: MSIE automatic proxy config

2000-08-27 Thread guy keren


On Sun, 27 Aug 2000, Shachar Shemesh wrote:

  maybe you should start thinking then ;) . if a "regular router" = cisco -
  then, yes, it can do that, and much more (depending on the version of its
  IOS).
 
 Maybe, but not as explained in your email.

actually, _exactly_ as explained in my email.

  this will done done with no address translation on the router - it just is
  told that the 'next hop' towards the target address,
 
 The "target address" is the entire internet. You are referring to the default
 route?

no. i think what i'm refering to falls under the specification of "policy
routing".

  is the proxy machine.
  that proxy machine then needs to understand (via normal routing rules)
  that any packet it received, targeted for port 80 and an IP that does not
  belogn to the local machine, should be injected into the proxy server's
  module. that doesn't _have_ to be implemented using NAT (althoguht it
  _might_ be done this way if it simplifies stuff).
 
 Yes, I agree. I have no problem with inplementing NAT on the proxy machine,
 BUT...

_if_ at all one needs NAT for that... or NAT in _any_ classical sense of
the word (according to your broad definitions, any using of a proxy server
is actually an introduction of NAT, since not the original machine's
addres is being shown in the FROM address of the packet being sent out,
but a different one (that of the proxy).

 You will find that your solution forwards ALL outbound packets to the proxy
 machine. Not just those aimed at port 80. 

actually, i won't. i'm talking of something that is actually used and
works as stated. i'm not sure how proficient you are with Cisco's IOS -
you might want to read their documentation before you state that this
cannot be done - because it is already being done. in fact, if one bothers
reading IOS's docs, one can do all sorts of non-standard things with their
routers.

 You are then left with my original
 problem - I don't want to penalise the entire office traffic with an extra hop
 (actually - extra two hops and a routing loop in your solution), just because
 I want to implement a transperant proxy. A much simpler solution for me is to
 block all communication to port 80 outbound, and force everyone to manually
 configure the proxy or they don't get web access.

simpler to whome exactly?

btw, please note that normally in our holy land, access bandwidth used to
a proxy server is MUCH MUCH smaller then the capacity of the LAN on which
this access is performed, so under common israely circumstances, this
waiste of resources is not realy an issue. surely, things are better if
all rowsers aer proeprly configured (less bandwidth waisted, about 1-3
milliseconds saved for each HTTP connection, and less router CPU cycles
are waisted) but sometimes it's easier and cheaper to support transparent
proxying in this way, then to support users with setting up the proxy
properly.

and since i think we're loosing our on-topicness by the minute here, i
think that if you still question Cisco's IOS features, we'll move this
discussion to private email.


guy

"For world domination - press 1,
 or dial 0, and please hold, for the creator." -- nob o. dy


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




kernel 2.2.16 and CD-ROM drives

2000-08-27 Thread Yosi

Hi,

I am using kernel 2.2.16 (plus patches to ResierFS and IDE). Recently
the kernel crashed while trying to rip audio track from a CD. Now,
If I remember correctly, 2.2.16 was issued after a serious security
bug was found, and I think that a short time after issuing 2.2.16,
Alan Cox issued a 2.2.17presomething that was supposed to fix a
problem with CD-ROM drives. Does anyone on the list remembers if that
was the case, and is it known that 2.2.16 has problems with some
brands of CD-ROM drives?

TIA,
Yosi

Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: MSIE automatic proxy config

2000-08-27 Thread Shachar Shemesh



guy keren wrote:

 _if_ at all one needs NAT for that... or NAT in _any_ classical sense of
 the word (according to your broad definitions, any using of a proxy server
 is actually an introduction of NAT, since not the original machine's
 addres is being shown in the FROM address of the packet being sent out,
 but a different one (that of the proxy).

 guy


Actually, NAT by the classical definition is any situation in which a "router"
modifies the packets it routes, but does leave the essence there. What you call
"NAT", is actually called "IP Masquarading", and is a particular instance of NAT.

A proxy is not a NAT. This is both because the packets are aimed at it (which is not
really the reason, as this holds true also of a transparent proxy), and because it
then initiates a totally different TCP connection with the real machine. Different
source IP, different TCP SYN numbering, different request. Everything is brand new.

As for the topic I was refering to at the begining, I will look up policy routing on
CISCO, even though it is not applicable to my particular case this time.

Shachar



=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: MSIE automatic proxy config

2000-08-27 Thread Gilad Ben-Yossef

Shachar Shemesh wrote:
 
 Gilad Ben-Yossef wrote:
 
  And once again I must say: "Don't think so 3rd layer, JeanLuke".
 
 I am not, number 1.

hehehe... I think in the moviwe it was the Borg Queen that said that ;-)

  I was about to explain how to build a 2d level (OSI) bridiging proxy but
  someone already did:
  http://perso.wanadoo.fr/magpie/EtherDivert.html
 
  No extra hop, no need for another subnet, batteries not included...
 
 Oh, but you do add an extra hop. The fact that no IP protocol is aware of that does
 not change the fact that you now require all your traffic to be directed through
 your box. The box still acts as a router (actually, a bridge, but same difference),
 and the performance penalties are still being payed (though I have to admit that
 it's probably less of a penalty).

That's not so accurate. It is an extra hop if you consider "a hop" every
piece of networking equipment  the packet (or Ethernet frame) passed on
it's merry way. But really - do you count the switches inside your LAN
as hops? for me a "hop" is really a router, something that decreases
TTL. A bridge is really not much then a repeater.

The work that is being done on a frame (which is not intended for the
Proxy) is much smaller with a bridge. In addition, you do not have to
"create" antoher subnet for the bridge, you don't have to to change the
router configuration for the bridge, you can replace the bridge with a
simply CAT5 cable in case of need and you can put a switch in paralel to
the bridge (giving it low STP priority) and get instant hot failover
solution without doing much.

In short - I understand why you say it is a hop, but the situation is
rather different from a "real" hop.

..which isn't quite relevant to the question wether you want to install
another machine or not just so the lusers can get their pr0n faster ;-)

-- 
Gilad Ben-Yossef [EMAIL PROTECTED] 
http://kagoor.com | +972(9)9565333 x230 | +972(54)756701
"I've been seduced by the chocolate side of the force."

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Syslog messages to a remote machine

2000-08-27 Thread Boaz Rymland

Hi folks,

I'm trying to set a server machine to send it's syslog messages to my machine.
For that I already managed to send all messages with :
*.* @my_machine

I've verified that indeed I get this on my machine with "tcpdump port 514" . I
can see messages poring in.
But, I cant seem to configure my local syslogd to do somethings with those.
I checked the documentation and found nothing about that. I also tried to add:
"in.syslog : the_remote_machines_IP" line in /etc/hosts.allow, but it didnt help
too.

Can you tip me on how to see (and log) those messages on the remote machine ?(my
machine).

TIA,
Boaz.

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: Syslog messages to a remote machine

2000-08-27 Thread Boaz Rymland

Boaz Rymland wrote:
 
 Hi folks,
 
 I'm trying to set a server machine to send it's syslog messages to my machine.
 For that I already managed to send all messages with :
 *.* @my_machine
 
 I've verified that indeed I get this on my machine with "tcpdump port 514" . I
 can see messages poring in.
 But, I cant seem to configure my local syslogd to do somethings with those.
 I checked the documentation and found nothing about that. I also tried to add:
 "in.syslog : the_remote_machines_IP" line in /etc/hosts.allow, but it didnt help
 too.
 
 Can you tip me on how to see (and log) those messages on the remote machine ?(my
 machine).
 
 TIA,
 Boaz.
 

Well, found it myself... ;-)

It's not in the configuration files but rather with a -r flag to the syslogd. On
a RH machine, edit /etc/rc.d/init.d/syslog file to add that flag there.

Boaz.

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




rpm

2000-08-27 Thread Emmanuel Lanzmann


I have Red Hat linux (6.1). I recently decided to upgrade rpm to 
3.0.5-9. Instead to do it with rpm, I decided to download the 
SOURCE (of rpm) and to compile it (I really like to see my computer
compiling !!). So far, evrything went great. 

Now, since then, when I try to install some package with rpm it 
almost always complains that a lot of stuff are missing (i.e.
dependencies). I am ready to admit that my libraries are certainly 
not all up to date (!!), but it often  complains for instance that

 /bin/sh is needed by  (the package I am installing)
 /usr/bin/perl is needed by 
 
Which I certainly do have !!!. 


So, something is apparently going wrong since I installed this new
version of rpm. Does someone has an idea ?

It is probably stupid, but I don't see it right now. 

Thanks in advance,

Emmanuel.



=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: Syslog messages to a remote machine

2000-08-27 Thread Ariel Biener

On Sun, 27 Aug 2000, Boaz Rymland wrote:


syslogd -r

RTFM.

--Ariel
 Hi folks,
 
 I'm trying to set a server machine to send it's syslog messages to my machine.
 For that I already managed to send all messages with :
 *.*   @my_machine
 
 I've verified that indeed I get this on my machine with "tcpdump port 514" . I
 can see messages poring in.
 But, I cant seem to configure my local syslogd to do somethings with those.
 I checked the documentation and found nothing about that. I also tried to add:
 "in.syslog : the_remote_machines_IP" line in /etc/hosts.allow, but it didnt help
 too.
 
 Can you tip me on how to see (and log) those messages on the remote machine ?(my
 machine).
 
 TIA,
 Boaz.
 
 =
 To unsubscribe, send mail to [EMAIL PROTECTED] with
 the word "unsubscribe" in the message body, e.g., run the command
 echo unsubscribe | mail [EMAIL PROTECTED]
 

--
Ariel Biener
e-mail: [EMAIL PROTECTED]   Work phone: 03-6406086
fingerprint = 07 D1 E5 3E EF 6D E5 82 0B E9 21 D4 3C 7D 8B BC


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: rpm

2000-08-27 Thread Gilad Ben-Yossef

Emmanuel Lanzmann wrote:
 
 I have Red Hat linux (6.1). I recently decided to upgrade rpm to
 3.0.5-9. Instead to do it with rpm, I decided to download the
 SOURCE (of rpm) and to compile it (I really like to see my computer
 compiling !!). So far, evrything went great.
 
 Now, since then, when I try to install some package with rpm it
 almost always complains that a lot of stuff are missing (i.e.
 dependencies). I am ready to admit that my libraries are certainly
 not all up to date (!!), but it often  complains for instance that
 
  /bin/sh is needed by  (the package I am installing)
  /usr/bin/perl is needed by 
 
 Which I certainly do have !!!.
 
 So, something is apparently going wrong since I installed this new
 version of rpm. Does someone has an idea ?
 

Yes, RPM keeps a database on things installed on the system. When you
recompiled from sources the installation proceess deleted the previous
RPM database and the RPM is not aware of anything that was installed
previously to it's update. Alternativly the RPm satabase was not deleted
but the new compiled RPM looks for it in a different place then the
default RPM that came with the system.

My suggestion? get a default RPM database from some random RH6.1
installation disk. I think it sits in a package of it's own or find
where the new RPm looks for that database and link to it.

Good luck,
Gilad.


-- 
Gilad Ben-Yossef [EMAIL PROTECTED] 
http://kagoor.com | +972(9)9565333 x230 | +972(54)756701
"I've been seduced by the chocolate side of the force."

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: Does anybody knows what happened to the php support of Apache?

2000-08-27 Thread Ilya Konstantinov

Omer Efraim wrote:
 
 Ilya Konstantinov wrote:
 
  Shlomi Fish wrote:
   Let me know when it is OK so I can add php support to Apache, and get the
   Links manager to run again.
 
  For some weird reason, both PHP3 and PHP4 keep segfaulting Apache on
  start.
  The gdb backtrace shows no hints.
  Two ideas come into mind:
  1) check the mysql shared libraries
  2) recompile Apache (weird Redhat stuff)
 
  I'll look into it later. Ideas, people?
  (meanwhile, PHP isn't there yet)
 
 DSO, right?
 
 Why don't you attach the backtrace anyhow.

It failed somewhere in the DSO loader. Seems to be nothing special.

Anyhow, I had a similar problem on a server we have here at work,
and finally I resolved it. For a start, PHP included their own
libmysqlclient with PHP4, probably since it's very popular and
many inexperienced users cannot figure out how to install
MySQL-shared. Unfortunatelly, it'll conflict and segfault if
anything else would access MySQL on the same server (be
DBI in mod_perl, or some mysql-based auth module). It's a known
bug and you can direct ./configure --with-mysql=/usr to link
against libmysqlclient.so and everything will work fine.

But still, it kept crashing with mod_perl loaded. Finally it
turned out DBD::mysql's mysql.so was linked against static
libmysqlclient.a. Relinking against the dynamic one solved the
problem. On my Debian desktop machine, everything went smoothly
since it's DBD::mysql comes dynamically-linked out of the box.

-- 
Best regards,
Ilya Konstantinov

=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: Syslog messages to a remote machine

2000-08-27 Thread guy keren


On Sun, 27 Aug 2000, Boaz Rymland wrote:

 I'm trying to set a server machine to send it's syslog messages to my machine.

just for general info: remote syslog is done by sending messages using
UDP, without any packet received acknowledgement or retransmission. thus,
if a logging packet is lost - the log message will be lost without any
sign for that. this is not recommended for a production system, if the
logs are important.

guy

"For world domination - press 1,
 or dial 0, and please hold, for the creator." -- nob o. dy


=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]




Re: rpm

2000-08-27 Thread Tzafrir Cohen

On Sun, 27 Aug 2000, Emmanuel Lanzmann wrote:

 
 I have Red Hat linux (6.1). I recently decided to upgrade rpm to
 3.0.5-9. Instead to do it with rpm, I decided to download the
 SOURCE (of rpm) and to compile it (I really like to see my computer
 compiling !!). So far, evrything went great.

Why did you choose rpm?

And why did you run "make install" if all you needed was to see your
computer compiling ? Anyway - something like 'rpm --rebuild
rpm-version.src.rpm' is the "right" way to compile your packages, and
still keep package management in tact.

 
 Now, since then, when I try to install some package with rpm it
 almost always complains that a lot of stuff are missing (i.e.
 dependencies). I am ready to admit that my libraries are certainly
 not all up to date (!!), but it oftencomplains for instance that
 
  /bin/sh is needed by  (the package I am installing)
  /usr/bin/perl is needed by 
 
 Which I certainly do have !!!.

There are two options:

1. your newly installed rpm uses a seperate database. The default rpm
database for redhat (probably other distros as well) is under /var/lib/rpm
 There are a couple of files with ".rpm" extention. Try to see what
database does rpm use (i.e. -run 'rpm -qf /bin/sh' and strace it). If it
uses a seperate database - see how to confgure it to run properly.

2. The installation overwrote the exiting database. You are in some truble
here, because you lost all the information about installed packages.
For the shortrun you may use the --nodep switchh to skip dependenccy
checks (don't use --fixed).
Anyway - I can't think of anything smarter than reinstallation. I don't
remeber any switch to rpm to update the database only. If such an option
exists - yu may manually rebuild your rpm database by adding all packages
into it. I think "rpm --rebuilddb" does something different.

 
 
 So, something is apparently going wrong since I installed this new
 version of rpm. Does someone has an idea ?
 
 It is probably stupid, but I don't see it right now.

I believe what happened is #2. to verify that - check if the install
script runs "rpm --initdb".

Good luck

-- 
Tzafrir Cohen
mailto:[EMAIL PROTECTED]
http://www.technion.ac.il/~tzafrir



=
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]