Ben Chitambira
2009-Feb-06 17:34 UTC
Fw: [Samba] specified network path no longer available
Following the problem below, I wanted to add that my users are accessing the samba shares using a single name \\Smb_Srv\share1 (which is set as the netbios name in smb.conf of all the nodes). Smb_Srv is set up in dns round robin load balancing to point to the ip addresses of the three nodes. I have just realised that if you make Smb_Srv point to only one fixed ip address, the clients don not get the error msg, although performance sucks. This is giving me another awkward idea: that the something is causing the clients to query the dns entry for Smb_Srv (more often and in the middle of the write) and when this happens, and the DNS returns another different IP (due to round robin), the client makes a new connection on this specific node, and closes the connection to the previos node, hence the error "specified network path bla bla.." If this really an issue for clustered samba or I am describing a false scenario? Does inreasing DNS TTL for Smb_Srv or these specific nodes help? Why do the windows clients have to query the dns within such a short space of time Thanks --- On Thu, 5/2/09, Ben Chitambira <bennychitambira@yahoo.com> wrote: From: Ben Chitambira <bennychitambira@yahoo.com> Subject: [Samba] specified network path no longer available To: samba@lists.samba.org Date: Thursday, 5 February, 2009, 3:40 PM I am having the "network path no longer available" issue and below is the log for the XP client concerned. This happens when copying a file from XP to the samba share. This is Centos 5, Sofs 1.5.2. with 3 nodes serving smb,nfs and ftp. domain controller is a separate linux samba pdc. auth is ldap. SO_RCVBUF and SO_SNDBUF both are 8192 with TCP_NODELAY. First technical consience tells me to set ?????? smb ports = 445 But the write_data failures are baffling. I am also considering increasing SO_RCVBUF and SNDBUFF How best can I address this? I believe I have the most recent/up to date software. This is affecting my users in production, so severity is critical. Thanks for help in advance. /var/log/samba/computer1.log [2009/02/05 12:05:29,? 0] modules/vfs_tsmsm.c:tsmsm_is_offline(209) ? Stale DMAPI session, re-creating it. [2009/02/05 12:05:30,? 1] smbd/service.c:close_cnum(1405) ? ls20094 (::ffff:10.31.6.30) closed connection to service share1 [2009/02/05 12:05:32,? 1] smbd/service.c:make_connection_snum(1194) ? ls20094 (::ffff:10.31.6.30) connect to service lmf initially as user bchitambira (uid=9426, gid=1010) (pid 31919) [2009/02/05 12:05:32,? 1] smbd/service.c:close_cnum(1405) ? ls20094 (::ffff:10.31.6.30) closed connection to service share1 [2009/02/05 12:23:53,? 0] lib/util_sock.c:write_data(1136) [2009/02/05 12:23:53,? 0] lib/util_sock.c:get_peer_addr_internal(1676) ? getpeername failed. Error was Transport endpoint is not connected ? write_data: write failure in writing to client 0.0.0.0. Error Connection reset by peer [2009/02/05 12:23:53,? 0] smbd/process.c:srv_send_smb(74) ? Error writing 4 bytes to client. -1. (Transport endpoint is not connected) [2009/02/05 12:24:04,? 0] lib/util_sock.c:write_data(1136) [2009/02/05 12:24:04,? 0] lib/util_sock.c:get_peer_addr_internal(1676) ? getpeername failed. Error was Transport endpoint is not connected ? write_data: write failure in writing to client 0.0.0.0. Error Connection reset by peer [2009/02/05 12:24:04,? 0] smbd/process.c:srv_send_smb(74) ? Error writing 4 bytes to client. -1. (Transport endpoint is not connected) Benny Chitambira *nixAdmin -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba
Ben Chitambira
2009-Feb-17 09:53 UTC
Fw: [Samba] specified network path no longer available
No joy upto now. Is there anything buggy in latest version of sofs-1.5.2? i.e samba-3.2.7-ctdb.54 --- On Fri, 6/2/09, Ben Chitambira <bennychitambira@yahoo.com> wrote: From: Ben Chitambira <bennychitambira@yahoo.com> Subject: Fw: [Samba] specified network path no longer available To: samba@lists.samba.org Date: Friday, 6 February, 2009, 5:34 PM Following the problem below, I wanted to add that my users are accessing the samba shares using a single name \\Smb_Srv\share1 (which is set as the netbios name in smb.conf of all the nodes). Smb_Srv is set up in dns round robin load balancing to point to the ip addresses of the three nodes. I have just realised that if you make Smb_Srv point to only one fixed ip address, the clients don not get the error msg, although performance sucks. This is giving me another awkward idea: that the something is causing the clients to query the dns entry for Smb_Srv (more often and in the middle of the write) and when this happens, and the DNS returns another different IP (due to round robin), the client makes a new connection on this specific node, and closes the connection to the previos node, hence the error "specified network path bla bla.." If this really an issue for clustered samba or I am describing a false scenario? Does inreasing DNS TTL for Smb_Srv or these specific nodes help? Why do the windows clients have to query the dns within such a short space of time Thanks --- On Thu, 5/2/09, Ben Chitambira <bennychitambira@yahoo.com> wrote: From: Ben Chitambira <bennychitambira@yahoo.com> Subject: [Samba] specified network path no longer available To: samba@lists.samba.org Date: Thursday, 5 February, 2009, 3:40 PM I am having the "network path no longer available" issue and below is the log for the XP client concerned. This happens when copying a file from XP to the samba share. This is Centos 5, Sofs 1.5.2. with 3 nodes serving smb,nfs and ftp. domain controller is a separate linux samba pdc. auth is ldap. SO_RCVBUF and SO_SNDBUF both are 8192 with TCP_NODELAY. First technical consience tells me to set ?????? smb ports = 445 But the write_data failures are baffling. I am also considering increasing SO_RCVBUF and SNDBUFF How best can I address this? I believe I have the most recent/up to date software. This is affecting my users in production, so severity is critical. Thanks for help in advance. /var/log/samba/computer1.log [2009/02/05 12:05:29,? 0] modules/vfs_tsmsm.c:tsmsm_is_offline(209) ? Stale DMAPI session, re-creating it. [2009/02/05 12:05:30,? 1] smbd/service.c:close_cnum(1405) ? ls20094 (::ffff:10.31.6.30) closed connection to service share1 [2009/02/05 12:05:32,? 1] smbd/service.c:make_connection_snum(1194) ? ls20094 (::ffff:10.31.6.30) connect to service lmf initially as user bchitambira (uid=9426, gid=1010) (pid 31919) [2009/02/05 12:05:32,? 1] smbd/service.c:close_cnum(1405) ? ls20094 (::ffff:10.31.6.30) closed connection to service share1 [2009/02/05 12:23:53,? 0] lib/util_sock.c:write_data(1136) [2009/02/05 12:23:53,? 0] lib/util_sock.c:get_peer_addr_internal(1676) ? getpeername failed. Error was Transport endpoint is not connected ? write_data: write failure in writing to client 0.0.0.0. Error Connection reset by peer [2009/02/05 12:23:53,? 0] smbd/process.c:srv_send_smb(74) ? Error writing 4 bytes to client. -1. (Transport endpoint is not connected) [2009/02/05 12:24:04,? 0] lib/util_sock.c:write_data(1136) [2009/02/05 12:24:04,? 0] lib/util_sock.c:get_peer_addr_internal(1676) ? getpeername failed. Error was Transport endpoint is not connected ? write_data: write failure in writing to client 0.0.0.0. Error Connection reset by peer [2009/02/05 12:24:04,? 0] smbd/process.c:srv_send_smb(74) ? Error writing 4 bytes to client. -1. (Transport endpoint is not connected) Benny Chitambira *nixAdmin -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba