OpenSSH and Postfix broken after Mandrake upgrade

My software firewall is disabled, as I am behind a router. My router is forwarding the ports correctly. I have restarted the services a number of times. When I try to SSH using puTTY, the black screen comes up for a minute, and then abruptly closes with no error message. What should be contained in my hosts.allow file to allow all connections?

For example your internal subnet IP is 192.168.0.

for /etc/hosts.allow

add

sshd 192.168.0.

This will only allow the internal subnet IP 192.168.0.x to connect to the server by ssh

Have you also upgarde the OpenSSH during your upgrade process? if not, try to do the upgrade of the ssh, download the lastest ssh rpm and issue the rpm -Uvh ssh.xxx.rpm. you better find the latest rpm from mandrake web site, if not you can try here ftp://ftp.openbsd.org/pub/OpenBSD/OpenSSH/portable/rpms/SRPMS/ and ftp://ftp.openbsd.org/pub/OpenBSD/OpenSSH/portable/rpm/SRPMS/

Have you also check the log files /var/log/messages and /var/log/secure, what's the message while you trying to connect by the puTTY?

As I said before, there is no error message when I try to use puTTY. The black screen comes up for a few moments, and then closes abruptly with no error message. I have also rebuilt OpenSSH from source, so I doubt that's the problem. What do I need to put in hosts.allow to allow ALL connections?

you said you have rebuild the ssh from source, have you copy the /openssh-version/contriv/redhat/sshd.init to /etc/rc.d/init.d ?

where you install the ssh? have you put the new version of sshd stuff into the /usr/bin?

since I'm now not in my office I can't show you which sshd stuff should put into the /usr/bin, but you can check those files and copy from the new ssh directory you install.

and the error messages, I mean have you check your linux server /var/log/messages and the /var/log/secure, is there any messages show about the sshd error?

Do you mean for everyone even outsider? if yes, it shoud be sshd:ALL will do, but not recommend, it might put you into the security risk.

Well, SSH just spontaneously started working today. paullamhkg: Although I didn't have to use any of your solutions, I will still give you some points after this question closes. However, I'm still having the problem with Postfix. If I do "telnet localhost 9265" to connect to the port Postfix is running on, I get a connection, but I don't see the message I should get that Postfix is running there. Also, when I do "netstat -tupan | grep 9265", the lines that come up are:

tcp        0      0 0.0.0.0:9265            0.0.0.0:*               LISTEN      29422/master
tcp        1      0 192.168.0.100:9265      62.65.69.4:55277        CLOSE_WAIT  -
tcp        1      0 127.0.0.1:9265          127.0.0.1:1533          CLOSE_WAIT  -
tcp        1      0 127.0.0.1:9265          127.0.0.1:1534          CLOSE_WAIT  -
tcp        1      0 192.168.0.100:9265      62.65.69.4:55239        CLOSE_WAIT  -

Since master is part of Postfix, I assume that first line is good, but I'm wondering if those four CLOSE_WAIT lines are part of the problem. Any help would be appreciated.

I should also note that although I'm using port 9265, I was still able to receive incoming mail because I have a computer that receives my e-mail on 25 and transfers it to 9265, then forwards it to this server.

Points is not important for me as long as you over come your problem, that's fine :)

The 4 CLOSE_WAIT 2 is for localhost(127.0.0.1) and 2 is 192.168.0.100 which I think is your IP of your linux box right :)

when your mail service start, then LISTEN, there will be a connection when receive mail, after mail delivered there should be close,  I think that's normal. But I don't understand why there is 2 closed, unless there is are 2 connections, I'm not using Postfix, I'm not quite sure, may be this is how it work, sorry can't provide info. for you, hope there is another EE can give you hints. :(

Paul

192.168.0.100 is indeed my network IP. 62.65.69.4 is the IP of the box that forwards my mail on 9265. How does one close a connection that is on CLOSE_WAIT? I had a similar problem once before with Apache which I fixed by rebooting, but I'm not particularly interested in doing that. Any thoughts?

May be that what we call a dead child, that's mean when a process completed and suppose it will finish all the process including closing, but it's wouldn't and left the process idle, the only way is kill the process or restart the service will do, no need to restart the linux box, but this only a temp solution, the solution is to check the config files to see any reason cause it and fix it, but as I said I'm not good in Postfix, can't give more detail for you.

The other thing which create the dead child is your mail client, I come acrossed in the past when I used fox mail as my client to connect to sendmail server, which always kept lots of dead child in my linux box, and I can't fix it, only thing I did was restart the sendmail service, the dead child gone. Even now, I'm using squirrellmail as my client, it's keep dead child, I've already report this to squirrelmail, hope they will give me the solution, because the dead child created by squirrelmail :(

Restarting the service doesn't make it go away... what is the syntax for using "kill" to get rid of it?

1st you have to know the process id, use ps -aux you will got the process id and kill -9 the (process id) will do.

Problem fixed... will request refund. Thanks for all of your help, paullamhkg.

No objections, but why not PAQ instead a delete?

Joeyman just wanted to learn how you fix your problem :)