RE: [Samba] Corrupted userid in mail folders - Crisis
Hi, another thing popes to my mind, check if a nscd process is running on your box. if yes stop it and remove it from the startup-scripts. It is not compatible with with samba and windbindd and may create strange effects. Christoph Yes, nscd was running, but I have stopped it now. What is nscd? Also, further to our discussion before, the userid actually get's changed Look at this: [EMAIL PROTECTED] root]# ls -l /home/RHENGHS/canhal total 12 drwx-- 7 monsla Domain Users 4096 Aug 17 13:48 Maildir/ drwx-- 2 monsla Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 monsla Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# ls -ln /home/RHENGHS/canhal total 12 drwx-- 7 10585 1 4096 Aug 17 13:48 Maildir/ drwx-- 2 10585 1 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 10585 1 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. Sep 22 07:49:05 inet nmbd[3724]: [2004/09/22 07:49:05, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:54:07 inet nmbd[3724]: [2004/09/22 07:54:07, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:28 inet smbd[9813]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:28 inet smbd[9813]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:28 inet smbd[9813]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:29 inet smbd[9815]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:29 inet smbd[9815]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:29 inet smbd[9815]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:56:02 inet nmbd[3724]: [2004/09/22 07:56:02, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:56:30 inet smbd[9832]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:write_socket(413) Sep 22 07:56:30 inet smbd[9832]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:send_smb(605) Sep 22 07:56:30 inet smbd[9832]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:56:53 inet smbd[9845]: [2004/09/22 07:56:53, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:53 inet smbd[9845]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:53 inet smbd[9845]: [2004/09/22 07:56:53, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:53 inet smbd[9845]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:53 inet smbd[9845]:
Re: [Samba] Corrupted userid in mail folders - Crisis
On Wednesday 22 September 2004 00:49, Roland Giesler wrote: Hi, another thing popes to my mind, check if a nscd process is running on your box. if yes stop it and remove it from the startup-scripts. It is not compatible with with samba and windbindd and may create strange effects. Christoph Yes, nscd was running, but I have stopped it now. What is nscd? The Name Service Caching Daemon. It can cause symptoms similar to what you are seeing. Do not run nscd with Samba, but also make certain that winbindd_idmap.tdb is not being deleted. - John T. Also, further to our discussion before, the userid actually get's changed Look at this: [EMAIL PROTECTED] root]# ls -l /home/RHENGHS/canhal total 12 drwx-- 7 monsla Domain Users 4096 Aug 17 13:48 Maildir/ drwx-- 2 monsla Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 monsla Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# ls -ln /home/RHENGHS/canhal total 12 drwx-- 7 10585 1 4096 Aug 17 13:48 Maildir/ drwx-- 2 10585 1 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 10585 1 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. Sep 22 07:49:05 inet nmbd[3724]: [2004/09/22 07:49:05, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:54:07 inet nmbd[3724]: [2004/09/22 07:54:07, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:28 inet smbd[9813]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:28 inet smbd[9813]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:28 inet smbd[9813]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:29 inet smbd[9815]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:29 inet smbd[9815]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:29 inet smbd[9815]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:56:02 inet nmbd[3724]: [2004/09/22 07:56:02, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:56:30 inet smbd[9832]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:write_socket(413) Sep 22 07:56:30 inet smbd[9832]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:send_smb(605) Sep 22 07:56:30 inet smbd[9832]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:56:53 inet smbd[9845]: [2004/09/22 07:56:53, 0]
RE: [Samba] Corrupted userid in mail folders - Crisis
A really silly question, but bear with me please... How do I remove a service from the cron.d directory? I can just delete the nscd file, but what if I want to run it again? Of course I can copy it somewhere into a backup dir to get it there again later, but it there is standard practice for disabling a service? Thanks Roland -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of John H Terpstra Sent: 22 September 2004 09:07 To: [EMAIL PROTECTED] Subject: Re: [Samba] Corrupted userid in mail folders - Crisis On Wednesday 22 September 2004 00:49, Roland Giesler wrote: Hi, another thing popes to my mind, check if a nscd process is running on your box. if yes stop it and remove it from the startup-scripts. It is not compatible with with samba and windbindd and may create strange effects. Christoph Yes, nscd was running, but I have stopped it now. What is nscd? The Name Service Caching Daemon. It can cause symptoms similar to what you are seeing. Do not run nscd with Samba, but also make certain that winbindd_idmap.tdb is not being deleted. - John T. Also, further to our discussion before, the userid actually get's changed Look at this: [EMAIL PROTECTED] root]# ls -l /home/RHENGHS/canhal total 12 drwx-- 7 monsla Domain Users 4096 Aug 17 13:48 Maildir/ drwx-- 2 monsla Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 monsla Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# ls -ln /home/RHENGHS/canhal total 12 drwx-- 7 10585 1 4096 Aug 17 13:48 Maildir/ drwx-- 2 10585 1 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 10585 1 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. Sep 22 07:49:05 inet nmbd[3724]: [2004/09/22 07:49:05, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:54:07 inet nmbd[3724]: [2004/09/22 07:54:07, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:28 inet smbd[9813]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:28 inet smbd[9813]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:28 inet smbd[9813]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:29 inet smbd[9815]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:29 inet smbd[9815]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:29 inet smbd[9815]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:56:02 inet nmbd[3724]: [2004/09/22 07:56:02, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:56:30 inet smbd[9832]: [2004/09/22 07:56:30, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:56:30 inet smbd[9832]: getpeername failed. Error was Transport
Re: [Samba] Corrupted userid in mail folders - Crisis
Hi again, Roland Giesler schrieb: Hi, another thing popes to my mind, check if a nscd process is running on your box. if yes stop it and remove it from the startup-scripts. It is not compatible with with samba and windbindd and may create strange effects. Christoph Yes, nscd was running, but I have stopped it now. What is nscd? Also, further to our discussion before, the userid actually get's changed Look at this: you didn't check what i wanted you to check... ;-) [EMAIL PROTECTED] root]# ls -l /home/RHENGHS/canhal total 12 drwx-- 7 monsla Domain Users 4096 Aug 17 13:48 Maildir/ drwx-- 2 monsla Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 monsla Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# ls -ln /home/RHENGHS/canhal total 12 drwx-- 7 10585 1 4096 Aug 17 13:48 Maildir/ drwx-- 2 10585 1 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 10585 1 4096 Jun 20 2002 webpage/ ^^^ what i want you to verify is if these numbers are changing. i bet they are constant and the only changing numbers are the numbers reported from getent. [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 No, i bet the number reported from getent is changing. Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. How was your samba installed? Was it compiled from source or are the package(s) from your distribution installed? Christoph -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
RE: [Samba] Corrupted userid in mail folders - Crisis
[EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 No, i bet the number reported from getent is changing. Ah, I understand what you mean. Let me check that. I need a predetermined mailbox to get corrupted... One that I know the userid number of. Ok, found one. You're right. The userid doesn't change, the mapping does [EMAIL PROTECTED] samba]# ls -l /home/RHENGHS/elahyl total 12 drwx-- 7 zelbeu Domain Users 4096 Aug 14 21:36 Maildir/ drwx-- 2 zelbeu Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 zelbeu Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] samba]# getent passwd zelbeu zelbeu:x:10865:1::/home/RHENGHS/zelbeu:/bin/bash [EMAIL PROTECTED] samba]# User elahyl is 1 - that I know. [EMAIL PROTECTED] samba]# getent passwd elahyl elahyl:x:1:1::/home/RHENGHS/elahyl:/bin/bash [EMAIL PROTECTED] samba]# I guess the userid on the folders is wrong now because I ran my fixing script. I'm going to run it again, since I change the slow hub to a switch and stopped the nscd service. The I'll see what happens. Roland Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. How was your samba installed? Was it compiled from source or are the package(s) from your distribution installed? Christoph -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
RE: [Samba] Corrupted userid in mail folders - Crisis
I've run the script that I posted earlier, and everything seems fine. I'll monitor the situation over the next hour or more to see what happens. The only effective change I made is that I stopped the nscd service. Roland -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Roland Giesler Sent: 22 September 2004 09:55 To: 'Christoph Scheeder' Cc: [EMAIL PROTECTED] Subject: RE: [Samba] Corrupted userid in mail folders - Crisis [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 No, i bet the number reported from getent is changing. Ah, I understand what you mean. Let me check that. I need a predetermined mailbox to get corrupted... One that I know the userid number of. Ok, found one. You're right. The userid doesn't change, the mapping does [EMAIL PROTECTED] samba]# ls -l /home/RHENGHS/elahyl total 12 drwx-- 7 zelbeu Domain Users 4096 Aug 14 21:36 Maildir/ drwx-- 2 zelbeu Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 zelbeu Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] samba]# getent passwd zelbeu zelbeu:x:10865:1::/home/RHENGHS/zelbeu:/bin/bash [EMAIL PROTECTED] samba]# User elahyl is 1 - that I know. [EMAIL PROTECTED] samba]# getent passwd elahyl elahyl:x:1:1::/home/RHENGHS/elahyl:/bin/bash [EMAIL PROTECTED] samba]# I guess the userid on the folders is wrong now because I ran my fixing script. I'm going to run it again, since I change the slow hub to a switch and stopped the nscd service. The I'll see what happens. Roland Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. How was your samba installed? Was it compiled from source or are the package(s) from your distribution installed? Christoph -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
Re: [Samba] Corrupted userid in mail folders - Crisis
On Wednesday 22 September 2004 01:18, Roland Giesler wrote: A really silly question, but bear with me please... How do I remove a service from the cron.d directory? I can just delete the nscd file, but what if I want to run it again? Of course I can copy it somewhere into a backup dir to get it there again later, but it there is standard practice for disabling a service? To stop a SuSE service: rc'servicename' stop ie: rcnscd stop For Red Hat Linux: /etc/rc.d/init.d/nscd stop Note: The path may be: /etc/init.d/'servicename' or /etc/rc.d/'servicename', or /etc/rc.d/init.d/'servicename' depending on your version and Linux vendor. To configure the service to prevent it from restarting when the server is rebooted: chkconfig nscd off To list all services and the status of each: chkconfig list Enjoy! - John T. Thanks Roland -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of John H Terpstra Sent: 22 September 2004 09:07 To: [EMAIL PROTECTED] Subject: Re: [Samba] Corrupted userid in mail folders - Crisis On Wednesday 22 September 2004 00:49, Roland Giesler wrote: Hi, another thing popes to my mind, check if a nscd process is running on your box. if yes stop it and remove it from the startup-scripts. It is not compatible with with samba and windbindd and may create strange effects. Christoph Yes, nscd was running, but I have stopped it now. What is nscd? The Name Service Caching Daemon. It can cause symptoms similar to what you are seeing. Do not run nscd with Samba, but also make certain that winbindd_idmap.tdb is not being deleted. - John T. Also, further to our discussion before, the userid actually get's changed Look at this: [EMAIL PROTECTED] root]# ls -l /home/RHENGHS/canhal total 12 drwx-- 7 monsla Domain Users 4096 Aug 17 13:48 Maildir/ drwx-- 2 monsla Domain Users 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 monsla Domain Users 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# ls -ln /home/RHENGHS/canhal total 12 drwx-- 7 10585 1 4096 Aug 17 13:48 Maildir/ drwx-- 2 10585 1 4096 Feb 23 2002 tmp/ drwxr-xr-x 2 10585 1 4096 Jun 20 2002 webpage/ [EMAIL PROTECTED] root]# getent passwd canhal canhal:x:10167:1::/home/RHENGHS/canhal:/bin/bash So you can see that the correct user should be 10167, not 10585 Also SMB hangs after a few hours. When I left last night, everything had been running fine for about 2 hours. When I got to work this morning, no-one was authenticated. When I tried to ls a user dir, I got not response. On issuing the reboot command I saw on the console that there was no process SMB. Here are the SMB messages in syslog. Sep 22 07:49:05 inet nmbd[3724]: [2004/09/22 07:49:05, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:54:07 inet nmbd[3724]: [2004/09/22 07:54:07, 0] libsmb/nmblib.c:send_udp(758) Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:28 inet smbd[9813]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:28 inet smbd[9813]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:28 inet smbd[9813]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection reset by peer Sep 22 07:55:28 inet smbd[9813]: [2004/09/22 07:55:28, 0] lib/util_sock.c:send_smb(605) Sep 22 07:55:28 inet smbd[9813]: Error writing 4 bytes to client. -1. (Connection reset by peer) Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:get_peer_addr(952) Sep 22 07:55:29 inet smbd[9815]: getpeername failed. Error was Transport endpoint is not connected Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket_data(388) Sep 22 07:55:29 inet smbd[9815]: write_socket_data: write failure. Error = Connection reset by peer Sep 22 07:55:29 inet smbd[9815]: [2004/09/22 07:55:29, 0] lib/util_sock.c:write_socket(413) Sep 22 07:55:29 inet smbd[9815]: write_socket: Error writing 4 bytes to socket 16: ERRNO = Connection
RE: [Samba] Corrupted userid in mail folders - Crisis
I run a network for a school as part of my duties and cannot resolve this problem. If I haven't given enough technical detail, please let me know, but I really need help with this. Thanks in advance Roland -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Roland Giesler Sent: 21 September 2004 14:48 To: [EMAIL PROTECTED] Subject: [Samba] Corrupted userid in mail folders I have a Mandrake 10.0 Official server running Samba3, Shorewall, Squid 2.5, Postfix and Courier-IMAP. Samba uses winbind to authenticate mail and proxy users against a windows 2000 ADS server. I get corruption happening in the user's home directories and elsewhere. The directory ownership changes all the time. One moment a dir belongs to roland:Domain Users and the next moment it's marjou:elahyl. The group and userid change, causing absolute havoc with mail delivery as wrong mail lands in people's mailboxes and users cannot be authenticated. My senior support technician is on honeymoon so I'm stuck. Restarting services makes no difference. If I run getent passwd username the results are 100% correct. Also for getend group groupname. Has anynone expereinced this or know of a fix. I'm somewhat clueless on some aspects, like I cannot check the version of Samba, since the doesn't appear to be a command switch for this. thanks Roland -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
RE: [Samba] Corrupted userid in mail folders - Crisis
Christoph Scheeder asked Hi, 1.) is this a new installation No, it was Mandrake 8.2 box, but was recently upgraded to Mandrake 10. 2.) if not what was changed when the failure first occured? Thing were running fine, at least that's what it appeared like. Then sporadically people started getting authentication error when logging into their IMAP mailboxes. Next the deputy principal reported that she had received other people's mail. So I ran the following script to reset all the folders: #!/bin/bash export IFS=: echo Auto-creating home directories for windows domain users getent passwd|grep :10[0-9][0-9][0-9]|while read USER FLAG USERUID USERGID USER_LONG USERHOME USERSHELL; do if [ -e $USERHOME ]; then echo Setting ownership of existing $USERHOME chown -R $USERUID.$USERGID $USERHOME else echo Creating $USERHOME for $USER with $USERUID/$USERGID mkdir $USERHOME; cp -a /etc/skel/* $USERHOME chown -R $USERUID.$USERGID $USERHOME fi done; After this the all appeared to be fine. But then a few hours later thing got even worse. Even squid's cache file got wrong ownership and stopped squid from spawning child processes. Eventually the machine hangs and I cannot see why. Btw is the a log where hanging processes can be logged? Or could I enable logging to syslog? I restarted the machine about an hour ago and all folders seem fine now 3.) please could you verify if the numeric user/group id's of the files in question change? use ls -ln on the files for this check. I would bet the numeric user/group id's don't change even when the usernames mapped to them change. please verify and post the result back to me. I'm checking that now. As soon as I have an id that's changed, I test and post the result. Roland -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba
Re: [Samba] Corrupted userid in mail folders - Crisis
On Tuesday 21 September 2004 10:20, Roland Giesler wrote: Christoph Scheeder asked Hi, 1.) is this a new installation No, it was Mandrake 8.2 box, but was recently upgraded to Mandrake 10. 2.) if not what was changed when the failure first occured? Thing were running fine, at least that's what it appeared like. Then sporadically people started getting authentication error when logging into their IMAP mailboxes. Next the deputy principal reported that she had received other people's mail. So I ran the following script to reset all the folders: Also, make sure that the file winbindd_idmap.tdb never gets deleted. It stores the SID to UID mappings and if it gets zapped you will see the symptoms you have reported. - John T. #!/bin/bash export IFS=: echo Auto-creating home directories for windows domain users getent passwd|grep :10[0-9][0-9][0-9]|while read USER FLAG USERUID USERGID USER_LONG USERHOME USERSHELL; do if [ -e $USERHOME ]; then echo Setting ownership of existing $USERHOME chown -R $USERUID.$USERGID $USERHOME else echo Creating $USERHOME for $USER with $USERUID/$USERGID mkdir $USERHOME; cp -a /etc/skel/* $USERHOME chown -R $USERUID.$USERGID $USERHOME fi done; After this the all appeared to be fine. But then a few hours later thing got even worse. Even squid's cache file got wrong ownership and stopped squid from spawning child processes. Eventually the machine hangs and I cannot see why. Btw is the a log where hanging processes can be logged? Or could I enable logging to syslog? I restarted the machine about an hour ago and all folders seem fine now 3.) please could you verify if the numeric user/group id's of the files in question change? use ls -ln on the files for this check. I would bet the numeric user/group id's don't change even when the usernames mapped to them change. please verify and post the result back to me. I'm checking that now. As soon as I have an id that's changed, I test and post the result. Roland -- John H Terpstra Samba-Team Member Phone: +1 (650) 580-8668 Author: The Official Samba-3 HOWTO Reference Guide, ISBN: 0131453556 Samba-3 by Example, ISBN: 0131472216 Hardening Linux, ISBN: 0072254971 OpenLDAP by Example, ISBN: 0131488732 Other books in production. -- To unsubscribe from this list go to the following URL and read the instructions: http://lists.samba.org/mailman/listinfo/samba