We run Samba 3.0.10 on Fedora Core 2. Every once in a while a file from photoshop can't be opened thru samba. Although it can be opened with the unix fs. When we gzip the file nothing is wrong and it can be send over the network again. Our designer has this problem on his machine, but when anybody else also tries to open the file they have the same problem. The log of samba says that the networkconnection has been resetted, but that is something that samba does (because only the file that locks has the problem and with all machines that try to open the file) Samba log says this: [2005/04/21 17:35:35, 2] smbd/open.c:open_file(245) bas opened file Projecten/CMS_3/Design/parts/splash_img.psd read=Yes write=No (numopen=8) [2005/04/21 17:35:35, 2] smbd/close.c:close_normal_file(270) nobody closed file Projecten/CMS_3/Design/parts/splash_img.psd (numopen=7) [2005/04/21 17:35:35, 2] smbd/open.c:open_file(245) bas opened file Projecten/CMS_3/Design/parts/splash_img.psd read=Yes write=No (numopen=8) [2005/04/21 17:35:35, 2] smbd/close.c:close_normal_file(270) nobody closed file Projecten/CMS_3/Design/parts/splash_img.psd (numopen=7) [2005/04/21 17:35:35, 2] smbd/open.c:open_file(245) bas opened file Projecten/CMS_3/Design/parts/splash_img.psd read=Yes write=No (numopen=8) [2005/04/21 17:36:52, 0] lib/util_sock.c:read_socket_data(384) read_socket_data: recv failure for 4. Error = Verbinding door partner opnieuw ingesteld [2005/04/21 17:36:52, 2] smbd/server.c:exit_server(571) Closing connections [2005/04/21 17:36:52, 1] smbd/service.c:close_cnum(836) bas2 (192.168.1.12) closed connection to service bas [2005/04/21 17:36:52, 1] smbd/service.c:close_cnum(836) bas2 (192.168.1.12) closed connection to service Data [2005/04/21 17:36:52, 2] smbd/close.c:close_normal_file(270) nobody closed file Projecten/CMS_3/Design/parts/splash_img.psd (numopen=7) [2005/04/21 17:36:52, 1] smbd/service.c:close_cnum(836) bas2 (192.168.1.12) closed connection to service Data [2005/04/21 17:36:52, 2] smbd/close.c:close_normal_file(270) bas closed file mail/cleanup.log (numopen=3) [2005/04/21 17:36:52, 2] smbd/close.c:close_normal_file(270) bas closed file mail/Pop3uidl.dbx (numopen=2) [2005/04/21 17:36:52, 2] smbd/close.c:close_normal_file(270) bas closed file mail/Folders.dbx (numopen=1) [2005/04/21 17:36:52, 2] smbd/close.c:close_normal_file(270) bas closed file mail/Offline.dbx (numopen=0) [2005/04/21 17:36:52, 1] smbd/service.c:close_cnum(836) bas2 (192.168.1.12) closed connection to service bas
I'm guessing this error means : read_socket_data: recv failure for 4. Error = Verbinding door partner opnieuw ingesteld the client closed the connection on us. This is usually due to network hardware problems. Can you translate this error message to English for me please ? Jeremy.
the error Verbinding door partner opnieuw ingesteld. means Connection reset by peer. And this isn't an hardware issue, if it was than only one pc should have the problem or the problem should also accure at other network connections at the same time. But if a file won't open it won't open an all computers that try to access it, and it only accures with photoshop files, though I also read reports on the internet that people have the same problem with word files. I am still trying to refind the forum I found the messages on. When they start yelling at me again about files that don't open i will upload one here for testing.
"Connection reset by peer" means the client just dropped the connection and sent a RST packet when we resent data. It either means a network hardware problem or a problem in the client TCP stack. There are *no* other options - that's what this error message means. There's nothing in the Samba code that is causing this. Jeremy.
if it was a hardware error it would not also occure on other computers and only with specific files. The files wouldn't be copyable with an nfs client either but they are. If it is a reset like you say, it is caused by the fact that samba locks up the line without finishing the conversation. The clients are all win2k machines and there is nothing wrong with their TCP stack. My guess is that it is caused by a binary combination that for some reason makes samba choke. Like i said when they start complaining about a file again i will post it here for verification.
"if it was a hardware error it would not also occure on other computers" - not neccessarily - it could be a switch/hub problem. "The files wouldn't be copyable with an nfs client either but they are." Also not true. NFS uses UDP (connectionless transport). SMB uses TCP and requires stream (continuous) connectivity. NFS survives bad networks much better than SMB. "If it is a reset like you say, it is caused by the fact that samba locks up the line without finishing the conversation." No, you don't understand. Samba is not "locking up". THE CLIENT IS DROPPING THE TCP CONNECTION. The important word here is ****"CLIENT"****. All Samba (the server) is doing is reporting to you that the client dropped the connection. We know nothing more. Jeremy.
closing. no response.