HP-UX

[FTP, TELNET] 질문을 다시 올립니다.

 
김은주
조언자

[FTP, TELNET] 질문을 다시 올립니다.

김병수님이 올려주신 답변이 좋았는데.. ( 패치부분,커널 수정등)

게시판 보다 모르구 삭제를 누르고 말았네요..ㅎㅎ

이런 실수를..

한번 더 올려주시면 정말 감사하겠습니다.



---------- 문제로그

...........

Mar 17 11:47:34 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:47:35 GIT ftpd: FTP session closed

Mar 17 11:48:43 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:48:44 GIT ftpd: FTP session closed

Mar 17 11:49:53 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:49:54 GIT ftpd: FTP session closed

Mar 17 11:50:55 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:50:55 GIT ftpd: FTP session closed

Mar 17 11:51:55 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:51:56 GIT ftpd: FTP session closed

Mar 17 11:52:55 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:52:56 GIT ftpd: FTP session closed

Mar 17 11:53:55 GIT telnetd: getpid: peer died: Error 0

Mar 17 11:53:56 GIT ftpd: FTP session closed

...........





~
1 응답 1
김병수
본과생

[FTP, TELNET] 질문을 다시 올립니다.

에구에구... 지울수도 있나요..그건 몰랐는데...



위의 error는 여러가지 경우로 인하여 발생될수 있습니다.



network ip 가 duplication 되어도 발생될수도 있습니다.



itrc에 있는 내용을 첨부합니다.



조치사항



1) Patch install :



PHNE_22159, PHNE_24762, PHNE_21835, PHNE_21767,PHNE_17434



2) ON 11.0 ---> 일반적으로 sam을 사용하기를 권장합니다...

You need to make some kernel changes, rebuild the kernel, and reboot your

machine to fix the error.

1. Change directories to /stand/build.

cd /stand/build

2. Create a new system file from the running kernel.

/usr/lbin/sysadm/system_prep -s system

3. Edit the kernel:

vi system



so that these parameters have these values:

npty 400

nstrpty 400

nstrtels 400

4. Build the new kernel.

/usr/sbin/mk_kernel -s system

5. Move the old system and vmunix files so if anything goes

wrong, the system can be booted from the old kernel.

mv /stand/system /stand/system.prev

mv /stand/vmunix /stand/vmunix.prev

mv /stand/dlkm /stand/dlkm.vmunix.prev

6. Move the new system file and new kernel into place, ready

to be used when rebooting the system:



mv /stand/build/system /stand/system

kmupdate /stand/build/vmunix_test

7. Reboot the system off the new kernel.

shutdown -r







3) First, check to see if rpcbind is running.

I have had problems in the past with this. Are you running dns?



If so, make sure your resolv.conf file is at least 444 permissions.



what does the entry for telnet in your inetd.conf file say?

have you tried re-reading in this file (inetd -c).

Can you give any other backround concerning the two machines in question?

Let me know what you find.



4) 이것은 itrc 에서 찾은 내용인데요..



누가 이런 글을 올려놨더라구요...한번 check 해보세요.



guppy inetd: Connection logging enabled

guppy inetd: telnet/tcp: Connection from unknown (65.160.118.4) at Wed Dec 5 11:14:26 2001

guppy telnetd: getpid: peer died: Error 0

guppy inetd: registrar/tcp: Connection from guppy (192.168.1.30) at Wed Dec 5 11:14:49 2001

guppy inetd: telnet/tcp: Connection from unknown (65.160.118.4) at Wed Dec 5 11:16:14 2001

guppy telnetd: getpid: peer died: Error 0



The Kbox (192.168.1.30) is located at another site behind a firewall. My machine at the office goes out on Cisco PIX firewall (65.160.118.4).



Could this be a firewall issue? My syslog.log is flooded with the peer died error.



Looks like the connections through the firewall are getting died.



firewall 을 check 해보라는 내용도 있습니다..



참조하시기 바랍니다..



위의 조건이 만족이 된상태라면,, 아래의 글을 참조하세요....



위의 error인 peer die가 발생할수 있는 경우를 전부 보냅니다.. 참조하시길 바랍니다...



<< PROBLEM >>



What is the meaning of the message telnetd getpid peer died error

0?





It means the tcp/ip connection to the telnet client has terminated

abnormally. RESOLUTION

Why would this message appear?



Usually there has been a communication failure between the telnetd process

and the telnet client process (the peer) running at the other side of the

tcp connection





Is this message a cause for alarm?

The occasional occurrence is not bad. For example, the telnet client may be

running on a PC that has hung, so the user simply powers off the PC.

In this case the telnetd process on HP-UX will detect the tcp connection

timeout and log the message.





If you want to proceed further with the investigation, you can identify the

client name or ip address that causes the message.



Enable verbose logging with #inetd -l. The telnet client ip address or

hostname is logged to /var/adm/syslog/syslog.log when the telnet client

makes the connection and inetd starts the telnetd pid. When the getpid error

is logged, check the pid number to see where the connection came from.



If the problem were insufficient npty, there would be an error message logged

by the telnetd pid in syslog: "Cannot allocate pty" and an error

message sent to the telnet client.





syslog.log file 안에 "cannot allocate pty"라는 구문이 없는 것으로 봐서 kernel을 늘리지 않아도 될듯합니다.



telnetd: ttloop: peer died: Connection timed out

... telnetd: recv: Connection reset by peer





RESOLUTION

Possible Solutions:



1. Are you telnetting in using hostname or ip? Do you get the same

delay when you telnet with hostname and/or ip?



a. If the delay is the same, look at /var/adm/nettl.LOG00 to see if this

log file got updated recently. If it was, this is an indication of a

duplciate IP address. Format the /var/adm/nettl.LOGXX file and look for an

indication of a duplicate IP in the output file:



# netfmt -Nlf /var/adm/nettl.LOG00 > /var/adm/nettl.out

# vi /var/adm/nettl.out

Pattern search on the string "dup" or "DUP"



b. If delay is longer telnet-ing in via hostname, look at name resolution

as the cause of this problem. Check for any DNS servers, as listed in

your /etc/resolv.conf file, as having named being unavailable. This may

be checked with nslookup:



# cat /etc/resolv.conf

domain rose.hp.com

nameserver 10.10.10.97

nameserver 10.10.10.98

nameserver 10.10.10.99



# nslookup

> server 10.10.10.97



Note: You may see a message such as:

"Specifying a server has overridden the switch policy order."

This is normal behavior.



> server 10.10.10.98



> server 10.10.10.99



Do this for the first 3 nameservers listed in the /etc/resolv.conf file.

To exit nslookup, enter a ^D.



If you find a nameserver that is not responding, try to find out if there

is a problem with the nameserver. Or, just remove that nameserver from the

/etc/resolv.conf file.



2. Drivers for PC NIC's:

What type of networkcards are in your PC's. We have had problems with certain

PC vendors network cards who had driver software which was out of date from

other PC's with the same type of cards. Download and install the most recent

version of drivers for the PC network cards you are using and see if the

problem clears up.



3. Other questions to answer or actions to take if the above steps do not

fix the problem. Take note of answers to these questions and the output from

these actions if necessary to open a Response Center Call.



a. Try to reset the card using lanadmin:

# lanadmin -> lan -> reset



b. What kind of NIC is the hp9000 server using?

Get and install the latest patches and dependancies for the network

interface hardware you have.



c. Is there a problem with the cable/hub/switch?

Get the checklist for 100BT troubleshooting problems. It can be found

in the ITRC as RCEN KBRC00001359.



d. How many clients are experiencing the same problem?



e. Are there any errors in dmesg output of note?



f. Is telnet fast for hpterms, but takes a long time with other emulators?



g. Does the hp9000 have a graphics console? If yes, is it slow in getting a

graphics login using CDE or VUE or X-Windows/Motif window manager?



h. On the PC, do you use ReflectionX or Exceed or some other type of terminal

emulator or are you doing XDMCP logins to the hp9000 server?



그럼 수고하세요...............................................................