Anonymous Servers - StartTLS with Mercury/32

Author

StartTLS with Mercury/32

Runaway Remailer Admin

2006-10-05, 7:13 am

This is a really frustrating experience. I tried now every alternative I
could think about, but I didn't get STARTTLS properly running with
Mercury/32. What I could set-up is secure SMTP via Stunnel on ports 465 and
587 but STARTTLS is not working via Stunnel (at least I was not able to find
a way to get it working) and if I switch SSL support on in Mercury/32 it
works a while and then it is stuck (some others report about the same
problem with STATRTLS and Mercury/32).
Is there any way to get STARTTLS working with Mercury or have I either just
to accept that there is no way with Mercury or swith to another MTA ?

Runaway Remailer Admin

madmax

2006-10-06, 1:17 am

On Thu, 5 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:
>This is a really frustrating experience.

that's certainly understandable

>I tried now every alternative I
>could think about, but I didn't get STARTTLS properly running with
>Mercury/32. What I could set-up is secure SMTP via Stunnel on ports 465 and
>587 but STARTTLS is not working via Stunnel (at least I was not able to find
>a way to get it working)

right, you can't do it, and setting
protocol = smtp
has no effect for server mode

>and if I switch SSL support on in Mercury/32 it
>works a while and then it is stuck (some others report about the same
>problem with STATRTLS and Mercury/32).
>Is there any way to get STARTTLS working with Mercury

since the project is dead as bikikii reports, then no

**BUT** there is a workaround: you use a server program which periodically
checks to see if your Mercury is behaving. (E.g., you can call openssl and
parse the output.) If Mercury is not providing SSL, then you auto restart
Mercury.

>or have I either just
>to accept that there is no way with Mercury

or just give up on having STARTTLS. Bikikii doesn't have it, Dingo doesn't
have it. I don't see it as such a big deal, EXCEPT that there are docs out
there that say you should/must use
protocol = smtp
in client stunnel configs.

>or swith to another MTA ?

I checked for you and saw that Inframail incorporates STARTTLS. I also
checked out Jakarta JAMES (a Java multi-platfrom MX) and saw that it has
smtps but not starttls.

IIRC Sendmail for Windows was free at one time. You might find old versions
at places like oldapps.org/com

>
>Runaway Remailer Admin

Runaway Remailer Admin

2006-10-06, 1:17 am

"madmax" <anon@comments.header> schrieb im Newsbeitrag
news:B9218YR538996.0247453704@twistycreek.com...
> On Thu, 5 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
> that's certainly understandable

>
>
> right, you can't do it, and setting
> protocol = smtp
> has no effect for server mode
>
>
> since the project is dead as bikikii reports, then no
>
> **BUT** there is a workaround: you use a server program which periodically
> checks to see if your Mercury is behaving. (E.g., you can call openssl and
> parse the output.) If Mercury is not providing SSL, then you auto restart
> Mercury.
>
>
> or just give up on having STARTTLS. Bikikii doesn't have it, Dingo doesn't
> have it. I don't see it as such a big deal, EXCEPT that there are docs out
> there that say you should/must use
> protocol = smtp
> in client stunnel configs.
>
>
>
> I checked for you and saw that Inframail incorporates STARTTLS. I also
> checked out Jakarta JAMES (a Java multi-platfrom MX) and saw that it has
> smtps but not starttls.
>
> IIRC Sendmail for Windows was free at one time. You might find old
> versions
> at places like oldapps.org/com
>
>
>

Thanks a lot for your useful explanations. I think at the moment I will give
up. I think my secure SMTP via Stunnel on ports 465 and 587 is now working
(as far as I can see). That allows users to connect via SSMTP on these
ports. So, there is no urgent need to get STARTTLS working. Mid term I will
think about your proposed workaround. May be there is already somebody who
wrote a "call openssl, parse the output and autostart mercury procedure" I
could re-use or adopt for my set-up" ???

Thanks again

Runaway Remailer Admin

madmax

2006-10-11, 7:12 pm

On Fri, 6 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:

[vbcol=seagreen]
>
>Thanks a lot for your useful explanations. I think at the moment I will give
>up. I think my secure SMTP via Stunnel on ports 465 and 587 is now working
>(as far as I can see). That allows users to connect via SSMTP on these
>ports. So, there is no urgent need to get STARTTLS working. Mid term I will
>think about your proposed workaround. May be there is already somebody who
>wrote a "call openssl, parse the output and autostart mercury procedure" I
>could re-use or adopt for my set-up" ???

can you run Java on your machine, Runaway? Would any other remop be
interested in a program to monitor Mercury and restart as needed? I could
write that when I have time, as a way to contribute to the cause here. Let
me know if anybody's interested.

Runaway Remailer Admin

2006-10-11, 7:12 pm

"madmax" <anon@comments.header> schrieb im Newsbeitrag
news:M6M4Z55Z39001.7462268519@twistycreek.com...
> On Fri, 6 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
>
>
>
> can you run Java on your machine, Runaway? Would any other remop be
> interested in a program to monitor Mercury and restart as needed? I could
> write that when I have time, as a way to contribute to the cause here. Let
> me know if anybody's interested.
>
>
>
I have Java Runtime Environment installed. May be Bikikii or twisty would
also be interested in a solution to shut down and restart Mercury dependent
on the STARTTLS status. I have read at least thatr twisty was also looking
for a solution to get STARTTLS working continously.

Runaway Admin.

BiKiKii

2006-10-11, 7:12 pm

madmax wrote in
news:M6M4Z55Z39001.7462268519@twistycreek.com:

> On Fri, 6 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
>

Mercury TLS is not good

This information posted in the past
Google groups or Merc mailing list

>

Maybe past post was not received...

465 is SMTPS (default)
587 is SMTP or STARTTLS Submission is not to be SMTPS

Your decision but... convention is something for adherence

> can you run Java on your machine, Runaway? Would any other remop be
> interested in a program to monitor Mercury and restart as needed? I
> could write that when I have time, as a way to contribute to the cause
> here. Let me know if anybody's interested.
>

There are many Mercury "Helpers" available at some sites

Example of interest:

<http://www.alahele.de/projects/>

MercupW v1.2e: mercury/32 watchdog

A windows tool to warn if Mercury/32 goes down

o sends e-mail via alternate server

o restarts Mercury/32 - immediately after crash and once every
15 minutes, if mercury/32 did crash again within that time period

o renames possibly broken queue files to avoid immediate recrash

o WILL PREVENT Windows from shutting down, while running

o PREVENTS the module windows of Mercury/32 from maximizing and closing

o can handle multiple instances of mercury/32

George Orwell

2006-10-12, 1:13 am

On 11 Oct 2006, madmax <anon@comments.header> wrote:
>On Fri, 6 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
>wrote:
>
>
>
>
>can you run Java on your machine, Runaway? Would any other remop be
>interested in a program to monitor Mercury and restart as needed? I could
>write that when I have time, as a way to contribute to the cause here. Let
>me know if anybody's interested.
>
>
>
>.

We are interested. Please code at your convienience.

Thanks

madmax

2006-10-13, 1:15 am

On Thu, 12 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:
>I have Java Runtime Environment installed. May be Bikikii or twisty would
>also be interested in a solution to shut down and restart Mercury dependent
>on the STARTTLS status. I have read at least thatr twisty was also looking
>for a solution to get STARTTLS working continously.

How do we handle the problem of you running code from an anonymous person?

If you'd like to compile it yourself, I will look up the link to J2SE JDK
for you.

Once installed, compiling is fairly easy.

madmax

2006-10-13, 1:15 am

On 11 Oct 2006, BiKiKii <anon@comments.header> wrote:

>There are many Mercury "Helpers" available at some sites
>
>
>Example of interest:
>
><http://www.alahele.de/projects/>
>
>
> MercupW v1.2e: mercury/32 watchdog
>
>A windows tool to warn if Mercury/32 goes down

thanks, bikikii. But we were discussing the situation where Mercury is stil
running, but is no longer doing TLS. Do you know if there is an exisitng
program for that?

Regardless, I had already spent an hour on it and got it half done, so I
went ahead and finished up a barebones program for now.

Do you know what happens when a running instance of Mercury no longer does
TLS? Does it no longer offer TLS, or does it offer TLS but fail to
negotiate it correctly?

>
> o sends e-mail via alternate server
>
> o restarts Mercury/32 - immediately after crash and once every
> 15 minutes, if mercury/32 did crash again within that time period
>
> o renames possibly broken queue files to avoid immediate recrash
>
> o WILL PREVENT Windows from shutting down, while running
>
> o PREVENTS the module windows of Mercury/32 from maximizing and closing
>
> o can handle multiple instances of mercury/32

madmax

2006-10-13, 1:15 am

madmax

2006-10-13, 1:15 am

-----BEGIN PGP SIGNED MESSAGE-----

On Thu, 12 Oct 2006, George Orwell <nobody@mixmaster.it> wrote:
>
>We are interested. Please code at your convienience.

Thanks for the feedback. Here's the base logic, for anybody that is
familiar with compiling Java. I tried to make it as simple as possible to
use. Feedback is appreciated.

This brings up the question of how to distribute this code? If I publish
the source, anyone can compile it but a person would need to have some
familiarity with doing that. Maybe some trusted person(s) could provide the
executable *.class file, or maybe crypto hashes can be used by others to
verify any class file that I publish, against the source code.

Paste into a file named MadMonitor.java
then run javac MadMonitor.java
to compile
then Java MadMonitor
c:\mercury\mercury.exe 25 30 to execute

=========================

/** MadMonitor by madmax@none.non
todo:
determine failure condition: does a Mercury instance in TLS failure
condition
1) give a 502 reply to the STARTTLS command?
or
2) accept the STARTTLS command but fail to negotiate TLS?

also, check if the target port (normally 25) is occupied before
starting Mercury

*/

import java.io.*;
import java.net.*;

public class MadMonitor {

private Runtime rt;
private Process proc;
private String exe;
private int port, interval;

public static void main(String[] args) {
if (args.length != 3) {
System.err.println("Usage: Java Madmonitor executable port
interval_seconds");
System.err.println("e.g.: Java Madmonitor
c:\\mercury\\mercury.exe 25 30");
} else {
new MadMonitor(args[0], args[1], args[2]).begin();
}
}

// class constructor
public MadMonitor(String exe, String port, String seconds) {
this.exe = exe;
this.port = Integer.parseInt(port);
interval = Integer.parseInt(seconds) * 1000;
System.out.println("Polling interval is " + seconds + " seconds.");
rt = Runtime.getRuntime();
}

private void begin() {
startProc();
boolean flag = true;
while (flag) {
try {
Thread.sleep(interval);

Socket sock = new Socket("127.0.0.1", port);

BufferedReader br = new BufferedReader
(new InputStreamReader(sock.getInputStream()));
PrintWriter pw = new PrintWriter
(new OutputStreamWriter(sock.getOutputStream()), true);

String s;
s = br.readLine(); // get greeting
System.out.println(s);

pw.println("STARTTLS");
s = br.readLine(); // get reply
System.out.println(s);

br.close();
pw.close();
sock.close();

if (s.startsWith("5")) {
endProc();
startProc();
}

} catch (InterruptedException ie) {
System.out.println(ie);
} catch (UnknownHostException uhe) {
System.err.println(uhe);
} catch (IOException ioe) {
System.err.println(ioe);
}
}
endProc();
}

private void endProc() {
proc.destroy();
try {
proc.waitFor();
System.out.print("Process destroyed. Exit value>" +
proc.exitValue());
System.out.println(" [Expected value=1]");
} catch (InterruptedException ie) {
System.err.println(ie);
}
}

private void startProc() {
try {
proc = rt.exec(exe);
System.out.println("MadMonitor started");
} catch (IOException ioe) {
System.out.println(ioe);
}
}

}

-----BEGIN PGP SIGNATURE-----
Version: N/A

iQA/AwUBRS8JUYc8LUQoEaFFEQKibQCg4r0dAZWM/r1eW6Snl/MCL09wpVQAmwX6
3PcesHt0QRf2td5/Y/RVTunP
=eQQl
-----END PGP SIGNATURE-----

madmax

2006-10-13, 1:15 am

-----BEGIN PGP SIGNED MESSAGE-----

On Thu, 12 Oct 2006, George Orwell <nobody@mixmaster.it> wrote:
>
>We are interested. Please code at your convienience.

Thanks for the feedback. Here's the base logic, for anybody that is
familiar with compiling Java. I tried to make it as simple as possible to
use. Feedback is appreciated.

This brings up the question of how to distribute this code? If I publish
the source, anyone can compile it but a person would need to have some
familiarity with doing that. Maybe some trusted person(s) could provide the
executable *.class file, or maybe crypto hashes can be used by others to
verify any class file that I publish, against the source code.

Paste into a file named MadMonitor.java
then run javac MadMonitor.java
to compile
then Java MadMonitor
c:\mercury\mercury.exe 25 30 to execute

=========================

/** MadMonitor by madmax@none.non
todo:
determine failure condition: does a Mercury instance in TLS failure
condition
1) give a 502 reply to the STARTTLS command?
or
2) accept the STARTTLS command but fail to negotiate TLS?

also, check if the target port (normally 25) is occupied before
starting Mercury

*/

import java.io.*;
import java.net.*;

public class MadMonitor {

private Runtime rt;
private Process proc;
private String exe;
private int port, interval;

public static void main(String[] args) {
if (args.length != 3) {
System.err.println("Usage: Java Madmonitor executable port
interval_seconds");
System.err.println("e.g.: Java Madmonitor
c:\\mercury\\mercury.exe 25 30");
} else {
new MadMonitor(args[0], args[1], args[2]).begin();
}
}

// class constructor
public MadMonitor(String exe, String port, String seconds) {
this.exe = exe;
this.port = Integer.parseInt(port);
interval = Integer.parseInt(seconds) * 1000;
System.out.println("Polling interval is " + seconds + " seconds.");
rt = Runtime.getRuntime();
}

private void begin() {
startProc();
boolean flag = true;
while (flag) {
try {
Thread.sleep(interval);

Socket sock = new Socket("127.0.0.1", port);

BufferedReader br = new BufferedReader
(new InputStreamReader(sock.getInputStream()));
PrintWriter pw = new PrintWriter
(new OutputStreamWriter(sock.getOutputStream()), true);

String s;
s = br.readLine(); // get greeting
System.out.println(s);

pw.println("STARTTLS");
s = br.readLine(); // get reply
System.out.println(s);

br.close();
pw.close();
sock.close();

if (s.startsWith("5")) {
endProc();
startProc();
}

} catch (InterruptedException ie) {
System.out.println(ie);
} catch (UnknownHostException uhe) {
System.err.println(uhe);
} catch (IOException ioe) {
System.err.println(ioe);
}
}
endProc();
}

private void endProc() {
proc.destroy();
try {
proc.waitFor();
System.out.print("Process destroyed. Exit value>" +
proc.exitValue());
System.out.println(" [Expected value=1]");
} catch (InterruptedException ie) {
System.err.println(ie);
}
}

private void startProc() {
try {
proc = rt.exec(exe);
System.out.println("MadMonitor started");
} catch (IOException ioe) {
System.out.println(ioe);
}
}

}

-----BEGIN PGP SIGNATURE-----
Version: N/A

iQA/ AwUBRS8UtIc8LUQoEaFFEQKOGwCbBphuKzFm0lBN
+32LTjq5dgDATaYAoP7+
FUGokG7ztWUpr7+Ep/uADUIQ
=gunD
-----END PGP SIGNATURE-----

Runaway Remailer Admin

2006-10-13, 1:15 am

"madmax" <madmax@none.non> schrieb im Newsbeitrag
news:F52MKNUB39002.9877430556@anonymous.poster...
> On 11 Oct 2006, BiKiKii <anon@comments.header> wrote:
>
>
> thanks, bikikii. But we were discussing the situation where Mercury is
> stil
> running, but is no longer doing TLS. Do you know if there is an exisitng
> program for that?
>
> Regardless, I had already spent an hour on it and got it half done, so I
> went ahead and finished up a barebones program for now.
>
> Do you know what happens when a running instance of Mercury no longer does
> TLS? Does it no longer offer TLS, or does it offer TLS but fail to
> negotiate it correctly?

As far as I could see looking at the Mercury SMTP Server it still offers
STARTTLS but it falis to negotiate it correctly. I can see how it offers on
my main MX and fails to negotiate and then the connection switches to my
backup MX. So, after a while running with STARTTLS mode it offers, but falis
to connect on the main MX, switches therefore the connection to the backup
MX without STARTTLS. That's how it looks like when I am looking at the
Mercury SMTP Server scrolling window. Hope this helps.

Runaway Admin

>
>
>
>
>
>
>
>
>
>
>

Runaway Remailer Admin

2006-10-13, 1:15 am

"madmax" <anon@comments.header> schrieb im Newsbeitrag
news:2S12O18339002.9904282407@twistycreek.com...
> On Thu, 12 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
> How do we handle the problem of you running code from an anonymous person?
>
> If you'd like to compile it yourself, I will look up the link to J2SE JDK
> for you.
>
> Once installed, compiling is fairly easy.
>

I have now installed the Windows Platform - J2SE(TM) Development Kit 5.0
Update 9
from the Sun Website. If you provide the source and a description how to
compile, I should be able to compile it by myself.

Runaway Admin

madmax

2006-10-13, 1:13 pm

On Fri, 13 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:

>
>As far as I could see looking at the Mercury SMTP Server it still offers
>STARTTLS but it falis to negotiate it correctly. I can see how it offers on
>my main MX and fails to negotiate and then the connection switches to my
>backup MX. So, after a while running with STARTTLS mode it offers, but falis
>to connect on the main MX, switches therefore the connection to the backup
>MX without STARTTLS. That's how it looks like when I am looking at the
>Mercury SMTP Server scrolling window. Hope this helps.
>
>Runaway Admin
>

Thanks, Runaway. That means I'd need a different scheme to detect TLS
failure. You can see in my code that I presently issue the STARTTLS
command, then check the first character of the server reply string, looking
for a "5".

I suppose I'll have to go back to my original notion of using OpenSSL, as
in stunnel, to check for TLS failure. That's for next week.

Were you able to compile and run the code? Java has to create the Mercury
subprocess in order for Java to be able to kill the Mercury subprocess.
What I did was simply to use my code to start Mercury, then use Mercury's
own smtp config dialog to turn off SSL., I then could observe that the Java
program behaved correctly (stopping Mercury then starting a new Mercury).

Can you give me an idea of what degree of use causes Mercury to stop doing
TLS? Is it a hundred msgs, a thousand, a mere dozen? I supose it's
difficult for you to know, because you wouldn't ordinarily know on which
port incoming msgs arrive.

Runaway Remailer Admin

2006-10-13, 7:17 pm

"madmax" <anon@comments.header> schrieb im Newsbeitrag
news:X4OUV4MJ39003.5725231481@twistycreek.com...
> On Fri, 13 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
>
> Thanks, Runaway. That means I'd need a different scheme to detect TLS
> failure. You can see in my code that I presently issue the STARTTLS
> command, then check the first character of the server reply string,
> looking
> for a "5".
>
> I suppose I'll have to go back to my original notion of using OpenSSL, as
> in stunnel, to check for TLS failure. That's for next week.
>
> Were you able to compile and run the code? Java has to create the Mercury
> subprocess in order for Java to be able to kill the Mercury subprocess.
> What I did was simply to use my code to start Mercury, then use Mercury's
> own smtp config dialog to turn off SSL., I then could observe that the
> Java
> program behaved correctly (stopping Mercury then starting a new Mercury).
>
> Can you give me an idea of what degree of use causes Mercury to stop doing
> TLS? Is it a hundred msgs, a thousand, a mere dozen? I supose it's
> difficult for you to know, because you wouldn't ordinarily know on which
> port incoming msgs arrive.
>

I tried compiling but I got the following error messages:

C:\Programme\Java\jdk1.5.0_09\bin>javac MadMonitor.java
MadMonitor.java:27: unclosed string literal
System.err.println("Usage: Java Madmonitor executable port
^
MadMonitor.java:28: unclosed string literal
interval_seconds");
^
MadMonitor.java:29: unclosed string literal
System.err.println("e.g.: Java Madmonitor
^
MadMonitor.java:30: illegal character: \92
c:\\mercury\\mercury.exe 25 30");
^
MadMonitor.java:30: illegal character: \92
c:\\mercury\\mercury.exe 25 30");
^
MadMonitor.java:30: illegal character: \92
c:\\mercury\\mercury.exe 25 30");
^
MadMonitor.java:30: illegal character: \92
c:\\mercury\\mercury.exe 25 30");
^
MadMonitor.java:30: unclosed string literal
c:\\mercury\\mercury.exe 25 30");
^
MadMonitor.java:31: ')' expected
} else {
^
9 errors

C:\Programme\Java\jdk1.5.0_09\bin>

Usually Mercury accepts about 30 messages with STARTTLS and then stops
connecting via TLS.

Runaway Admin

madmax

2006-10-13, 7:17 pm

sorry, Runaway. The errors resulted from the wrapping of the source that
occurred during posting here.

The specific problem is with the following two lines of source. The lines
are:

System.err.println("Usage: Java Madmonitor executable port
interval_seconds");
System.err.println("e.g.: Java Madmonitor
c:\\mercury\\mercury.exe 25 30");

If you repair the source so that each of those two lines is no longer
split, then all will work well. In other words, the begin-quote and
end-quote need to be on the same line.

(I'd guess that the M2N did the line wrapping.)

Also, you should be able to use any arbitrary directory to work in, since I
believe the installation of Java will set the OS path accordingly, to the
<Java Home>/bin directory.

On Fri, 13 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:
>
>I tried compiling but I got the following error messages:
>
>C:\Programme\Java\jdk1.5.0_09\bin>javac MadMonitor.java
>MadMonitor.java:27: unclosed string literal
> System.err.println("Usage: Java Madmonitor executable port
> ^
>MadMonitor.java:28: unclosed string literal
>interval_seconds");
> ^
>MadMonitor.java:29: unclosed string literal
> System.err.println("e.g.: Java Madmonitor

Runaway Remailer Admin

2006-10-13, 7:17 pm

"madmax" <madmax@none.non> schrieb im Newsbeitrag
news:WHC63YB039003.6424305556@anonymous.poster...
> sorry, Runaway. The errors resulted from the wrapping of the source that
> occurred during posting here.
>
> The specific problem is with the following two lines of source. The lines
> are:
>
> System.err.println("Usage: Java Madmonitor executable port
> interval_seconds");
> System.err.println("e.g.: Java Madmonitor
> c:\\mercury\\mercury.exe 25 30");
>
> If you repair the source so that each of those two lines is no longer
> split, then all will work well. In other words, the begin-quote and
> end-quote need to be on the same line.
>
> (I'd guess that the M2N did the line wrapping.)
>
> Also, you should be able to use any arbitrary directory to work in, since
> I
> believe the installation of Java will set the OS path accordingly, to the
> <Java Home>/bin directory.
>
>
> On Fri, 13 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
>

Thanks a lot, I was able to compile it now and started Mercury via the
compiled java. Let's see whether STARTTLS works now. Can you test it may be
? I have just switched Mercury via MadMonitor on and also Reliable.

Runaway Remailer Admin

2006-10-15, 7:13 am

"Runaway Remailer Admin" <anonymous@nym.alias.net> schrieb im Newsbeitrag
news:egp19c$kfq$1@bananasplit.info...
>
> "madmax" <madmax@none.non> schrieb im Newsbeitrag
> news:WHC63YB039003.6424305556@anonymous.poster...
>
> Thanks a lot, I was able to compile it now and started Mercury via the
> compiled java. Let's see whether STARTTLS works now. Can you test it may
> be ? I have just switched Mercury via MadMonitor on and also Reliable.
>

I have watched now the STARTTLS funcionality (using the compiled java) the
last 24 hours. I looked at the Mercury SMTP Server scrolling window and
tested it with openssl s-client ... command. My observations where:

- I think with your Java program it was now running longer with correct
STARTTLS functionality
- Unfortunately STARTTLS functionality in Mercury is still crashing after a
while (sometimesafter an hour, sometimes after 4 hours). When I shut down
Mercury and restart it manually using your Java program STARTTLS
functionality is working again.
- I think that that just after a while running properly it STARTTLS does not
connect anymore properly and only a shut-down and re-start of Mercury get's
STARTTLS functionality re-instated. May be if you could add a error check in
your Java checking the STARTTLS connectivity and shutting down Mercury and
restart it, we could create a workaround for this Mercury STARTTLS bug. What
I could see was that sometimes it was required to shut-down and restart
Mercury about three times until STARTTLS came back woking properly. So, the
error check would need to be done also directly after starting mercury via
your Java to make sure that it is restarted several times in case one
restart will not be sufficient to get back STARTTLS working properly. I hope
that my observations are of help. I don't know whether this could be done by
you adopting the current Java but if, it is possible, I certainly would
compile an adopted version a test it. Thanks for all your efforts.

Runaway Admin

madmax

2006-10-19, 7:12 pm

On Sun, 15 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
wrote:

>- I think with your Java program it was now running longer with correct
>STARTTLS functionality
>- Unfortunately STARTTLS functionality in Mercury is still crashing after a
>while (sometimesafter an hour, sometimes after 4 hours). When I shut down
>Mercury and restart it manually using your Java program STARTTLS
>functionality is working again.

Hi, Runaway. The manner with which Mercury is started shouldn't affect how
it runs - unless you use some switch that I'm not aware of.

>- I think that that just after a while running properly it STARTTLS does not
>connect anymore properly and only a shut-down and re-start of Mercury get's
>STARTTLS functionality re-instated. May be if you could add a error check in
>your Java checking the STARTTLS connectivity and shutting down Mercury and
>restart it, we could create a workaround for this Mercury STARTTLS bug.

Yes, indeed. Sorry I have been sidetracked this week, but I do intend to
get to it.

> What
>I could see was that sometimes it was required to shut-down and restart
>Mercury about three times until STARTTLS came back woking properly.

That sure sounds odd.

>So, the
>error check would need to be done also directly after starting mercury via
>your Java to make sure that it is restarted several times in case one
>restart will not be sufficient to get back STARTTLS working properly.

Okay. This all might not even be worth it, but we might as well keep going
and see it through to completion. What do you think?

> I hope
>that my observations are of help. I don't know whether this could be done by
>you adopting the current Java but if, it is possible, I certainly would
>compile an adopted version a test it. Thanks for all your efforts.

and thank you, especially for running a remailer in the first place.

>
>Runaway Admin

Runaway Remailer Admin

2006-10-19, 7:12 pm

"madmax" <anon@comments.header> schrieb im Newsbeitrag
news:9LDWP8SE39009.6618981482@twistycreek.com...
> On Sun, 15 Oct 2006, "Runaway Remailer Admin" <anonymous@nym.alias.net>
> wrote:
>
>
> Hi, Runaway. The manner with which Mercury is started shouldn't affect how
> it runs - unless you use some switch that I'm not aware of.
>
>
> Yes, indeed. Sorry I have been sidetracked this week, but I do intend to
> get to it.
>
>
> That sure sounds odd.
>
>
> Okay. This all might not even be worth it, but we might as well keep going
> and see it through to completion. What do you think?

I think we should give it a try. I will at least test to get it running. If
it fails, so what. If we get it running may be other Remops can / want also
make also use of it.
Let me thank you for your Java contributions and efforts

Runaway Remailer Admin
>
>
> and thank you, especially for running a remailer in the first place.
>
>
>