Ramsey Wes - wramse
2006-Aug-23 12:11 UTC
[Samba] Problem with large files corrupting during transfer
I'm very new to Samba, and I'm supporting my team's migration to a new server. We are running v3.0.23b, ML3 and AIX v5.3, and we're pulling data from NT servers. V2.2.2 (what we are using on the old server) would not compile on the new box, so we have to upgrade. Files >2.8gb are corrupting during the transfer. That 2.8gb file is the largest we've gotten to go through successfully. The next largest file is 5.7gb, and corrupts without fail. The 1st section of the file looks normal, the next large chunk is corrupt, and the remainder of the file is simply binary trash. It looks to me like 2(+) users are accessing the file while it is writing to cache, which would explain this, but I have no way to prove it. I tried adding 'oplocks = no' and 'strict locking yes', but neither fixed the problem. If more info is needed, please let me know. Here's the smb.conf: [global] workgroup = CORP server string = Qbert at Samba 3.0.23b interfaces = xxx.xx.xxx.xxx security = DOMAIN password server = * map to guest = Bad User log level = 2 log file = /usr/local/samba/var/log.%m log file = /usr/local/samba/var/log.%S max log size = 10000 deadtime = 150 max open files = 2048 local master = No domain master = No dns proxy = No wins server = 10.87.225.225 create mask = 0774 directory mask = 0775 oplocks = No socket options = TCP_NODELAY Here's the log.%S: [2006/08/22 18:00:44, 2] smbd/reply.c:reply_special(496) netbios connect: name1=QBERT name2=DPREP3 [2006/08/22 18:00:44, 2] smbd/reply.c:reply_special(503) netbios connect: local=qbert remote=dprep3, name type = 0 [2006/08/22 18:00:44, 2] smbd/sesssetup.c:setup_new_vc_session(795) setup_new_vc_session: New VC == 0, if NT4.x compatible we would close all old resources. [2006/08/22 18:00:44, 2] libsmb/namequery.c:name_query(577) Got a positive name query response from 10.87.225.225 ( xx.xx.xxx.xxx xx.xx.xx.xxx xx.xx.xxx.xxx xx.xx.xxx.xxx xx.xx.xx.xxx... [2006/08/22 18:00:44, 2] libsmb/namequery.c:name_query(577) Got a positive name query response from 10.87.225.225 ( 10.87.225.225 ) [2006/08/22 18:00:53, 1] lib/util_sock.c:open_socket_out(896) timeout connecting to 10.87.225.225:445 [2006/08/22 18:01:02, 1] lib/util_sock.c:open_socket_out(896) timeout connecting to 10.87.225.225:445 [2006/08/22 18:01:02, 2] auth/auth.c:check_ntlm_password(309) check_ntlm_password: authentication for user [NAVSVR2] -> [NAVSVR2] -> [navsvr2] succeeded [2006/08/22 18:01:02, 2] smbd/reply.c:reply_tcon_and_X(711) Serving IPC$ as a Dfs root [2006/08/22 18:01:02, 1] smbd/service.c:make_connection_snum(941) dprep3 (139.61.176.211) connect to service dpspot initially as user navsvr2 (uid=4126, gid=209) (pid 283206) [2006/08/22 18:01:02, 2] smbd/reply.c:reply_tcon_and_X(711) Serving dpspot as a Dfs root [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=Yes (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=No (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=No (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) [2006/08/22 18:21:16, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/per.file1 (numopen=0) [2006/08/22 18:21:16, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/per.file1 read=No write=No (numopen=1) [2006/08/22 18:21:16, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/per.cust1 (numopen=0) [2006/08/22 18:31:54, 1] smbd/service.c:close_cnum(1141) dprep3 (139.61.176.211) closed connection to service dpspot Here's the log.nmbd: (This exact message is repeated every 5 minutes) [2006/08/23 06:46:14, 2] nmbd/nmbd_elections.c:send_election_dgram(41) send_election_dgram: Sending election packet for workgroup CORP on subnet xxx.xx.xxx.xxx *************************************************************************** The information contained in this communication is confidential, is intended only for the use of the recipient named above, and may be legally privileged. If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please resend this communication to the sender and delete the original message or any copy of it from your computer system. Thank You. ****************************************************************************
Ramsey Wes - wramse
2006-Aug-24 13:47 UTC
[Samba] RE: Problem with large files corrupting during transfer
List: I've gotten more opinions on this problem, and the consensus is that Samba is not compatible with Windows 2000 server (or NT) and our Unix environment. If this is true I'm wasting my time. Is this a common or known bug, or is it just something in the .conf files that needs to be setup differently? ________________________________ From: Ramsey Wes - wramse Sent: Wednesday, August 23, 2006 7:01 AM To: 'samba@lists.samba.org' Subject: Problem with large files corrupting during transfer I'm very new to Samba, and I'm supporting my team's migration to a new server. We are running v3.0.23b, ML3 and AIX v5.3, and we're pulling data from NT servers. V2.2.2 (what we are using on the old server) would not compile on the new box, so we have to upgrade. Files >2.8gb are corrupting during the transfer. That 2.8gb file is the largest we've gotten to go through successfully. The next largest file is 5.7gb, and corrupts without fail. The 1st section of the file looks normal, the next large chunk is corrupt, and the remainder of the file is simply binary trash. It looks to me like 2(+) users are accessing the file while it is writing to cache, which would explain this, but I have no way to prove it. I tried adding 'oplocks = no' and 'strict locking yes', but neither fixed the problem. If more info is needed, please let me know. Here's the smb.conf: [global] workgroup = CORP server string = Qbert at Samba 3.0.23b interfaces = xxx.xx.xxx.xxx security = DOMAIN password server = * map to guest = Bad User log level = 2 log file = /usr/local/samba/var/log.%m log file = /usr/local/samba/var/log.%S max log size = 10000 deadtime = 150 max open files = 2048 local master = No domain master = No dns proxy = No wins server = 10.87.225.225 create mask = 0774 directory mask = 0775 oplocks = No socket options = TCP_NODELAY Here's the log.%S: [2006/08/22 18:00:44, 2] smbd/reply.c:reply_special(496) netbios connect: name1=QBERT name2=DPREP3 [2006/08/22 18:00:44, 2] smbd/reply.c:reply_special(503) netbios connect: local=qbert remote=dprep3, name type = 0 [2006/08/22 18:00:44, 2] smbd/sesssetup.c:setup_new_vc_session(795) setup_new_vc_session: New VC == 0, if NT4.x compatible we would close all old resources. [2006/08/22 18:00:44, 2] libsmb/namequery.c:name_query(577) Got a positive name query response from 10.87.225.225 ( xx.xx.xxx.xxx xx.xx.xx.xxx xx.xx.xxx.xxx xx.xx.xxx.xxx xx.xx.xx.xxx... [2006/08/22 18:00:44, 2] libsmb/namequery.c:name_query(577) Got a positive name query response from 10.87.225.225 ( 10.87.225.225 ) [2006/08/22 18:00:53, 1] lib/util_sock.c:open_socket_out(896) timeout connecting to 10.87.225.225:445 [2006/08/22 18:01:02, 1] lib/util_sock.c:open_socket_out(896) timeout connecting to 10.87.225.225:445 [2006/08/22 18:01:02, 2] auth/auth.c:check_ntlm_password(309) check_ntlm_password: authentication for user [NAVSVR2] -> [NAVSVR2] -> [navsvr2] succeeded [2006/08/22 18:01:02, 2] smbd/reply.c:reply_tcon_and_X(711) Serving IPC$ as a Dfs root [2006/08/22 18:01:02, 1] smbd/service.c:make_connection_snum(941) dprep3 (139.61.176.211) connect to service dpspot initially as user navsvr2 (uid=4126, gid=209) (pid 283206) [2006/08/22 18:01:02, 2] smbd/reply.c:reply_tcon_and_X(711) Serving dpspot as a Dfs root [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=Yes (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=No (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/TEST92827.TXT read=No write=No (numopen=1) [2006/08/22 18:01:02, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/TEST92827.TXT (numopen=0) [2006/08/22 18:01:02, 2] smbd/open.c:open_file(352) [2006/08/22 18:21:16, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/per.file1 (numopen=0) [2006/08/22 18:21:16, 2] smbd/open.c:open_file(352) NAVSVR2 opened file new.cust/per.file1 read=No write=No (numopen=1) [2006/08/22 18:21:16, 2] smbd/close.c:close_normal_file(344) navsvr2 closed file new.cust/per.cust1 (numopen=0) [2006/08/22 18:31:54, 1] smbd/service.c:close_cnum(1141) dprep3 (139.61.176.211) closed connection to service dpspot Here's the log.nmbd: (This exact message is repeated every 5 minutes) [2006/08/23 06:46:14, 2] nmbd/nmbd_elections.c:send_election_dgram(41) send_election_dgram: Sending election packet for workgroup CORP on subnet xxx.xx.xxx.xxx *************************************************************************** The information contained in this communication is confidential, is intended only for the use of the recipient named above, and may be legally privileged. If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please resend this communication to the sender and delete the original message or any copy of it from your computer system. Thank You. ****************************************************************************
Ramsey Wes - wramse
2006-Aug-24 17:36 UTC
[Samba] RE: Problem with large files corrupting during transfer
Jeremy, We have an automated process that pulls data files from our server (Qbert specifically) - by the way, ftp up is fine and files are moving just fine. When the process/program is done manipulating the data it writes it out from the NT or Win2k Server server directly to Qbert (so that the data is not stored any place other than the destination server) in 3 different file segments, using userid 'navsvr2'. The smaller two of the 3 files are fine (<2gb), but the largest (5.7gb) gets corrupted. If I can provide any other information, please let me know. -----Original Message----- From: Jeremy Allison [mailto:jra@samba.org] Sent: Thursday, August 24, 2006 10:53 AM To: Ramsey Wes - wramse Cc: samba@lists.samba.org Subject: Re: [Samba] RE: Problem with large files corrupting during transfer On Thu, Aug 24, 2006 at 08:46:02AM -0500, Ramsey Wes - wramse wrote:> > I've gotten more opinions on this problem, and the consensus is that > Samba is not compatible with Windows 2000 server (or NT) and our Unix > environment. If this is true I'm wasting my time. Is this a common or > known bug, or is it just something in the .conf files that needs to be > setup differently?No, that's not true.> server. We are running v3.0.23b, ML3 and AIX v5.3, and we're pulling > data from NT servers. V2.2.2 (what we are using on the old server)would> not compile on the new box, so we have to upgrade.How are you pulling the data ? Jeremy. *************************************************************************** The information contained in this communication is confidential, is intended only for the use of the recipient named above, and may be legally privileged. If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please resend this communication to the sender and delete the original message or any copy of it from your computer system. Thank You. ****************************************************************************
James Watkins
2006-Aug-25 08:27 UTC
[Samba] Problem with large files corrupting during transfer
Ramsey Wes - wramse wrote:> Files >2.8gb are corrupting during the transfer. That 2.8gb file is the > largest we've gotten to go through successfully. The next largest file > is 5.7gb, and corrupts without fail. The 1st section of the file looks > normal, the next large chunk is corrupt, and the remainder of the file > is simply binary trash. It looks to me like 2(+) users are accessing the > file while it is writing to cache, which would explain this, but I have > no way to prove it. I tried adding 'oplocks = no' and 'strict locking > yes', but neither fixed the problem. If more info is needed, please let > me know.Hi, I'm afraid I can't help you but I can tell you that I had a similar problem that I gave up on some time ago. I found that windows 2k could transfer the files perfectly well to my samba box so I guess it must be an NT problem. I posted this to the list: http://lists.samba.org/archive/samba/2006-May/120717.html and only received one reply: http://lists.samba.org/archive/samba/2006-May/120748.html Having verified that I was running SP6, I lost interest and tried to persuade my boss to fund an upgrade to our windows servers instead. I'm assured that the upgrade will happen some time this year :-) but I'd still be very interested to hear from you if you manage to fix the problem. Good luck, James.
Ramsey Wes - wramse
2006-Aug-25 12:58 UTC
[Samba] Problem with large files corrupting during transfer
That's not exactly what I wanted to hear, but I guess we have no choice but to build a work around. Our NT servers are no longer supported because of an upgrade to a new system, and we're one of the few teams left using the old system - therefore, I can't request any patches or changes be applied on that end. Our migration has been discussed, but it will take several months and I'm afraid we're already weeks behind going to this new server. One other option though...is there an older version of Samba that might work between NT and our AIX box? We tried yesterday to install 2.2.8a, and this is the return: Compiling printing/nt_printing.c Compiling smbd/noquotas.c Compiling smbd/oplock.c Compiling smbd/oplock_irix.c Compiling smbd/oplock_linux.c Compiling smbd/notify.c Compiling smbd/notify_hash.c Compiling smbd/notify_kernel.c Linking bin/smbd ld: 0711-317 ERROR: Undefined symbol: .VA_COPY ld: 0711-317 ERROR: Undefined symbol: .SAFE_FREE ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more information. collect2: ld returned 8 exit status make: 1254-004 The error code from the last command is 1. Is it possible to fix the makefile so that it will work? *************************************************************************** The information contained in this communication is confidential, is intended only for the use of the recipient named above, and may be legally privileged. If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please resend this communication to the sender and delete the original message or any copy of it from your computer system. Thank You. ****************************************************************************