From Bugzilla Helper: User-Agent: Mozilla/5.0 (compatible; Konqueror/3.5; Linux) KHTML/3.5.8 (like Gecko) Description of problem: After upgrading to EL-5.1 (i386), some XP sp2 clients report intermittently "Delayed write failed" error when writing to the samba share, read access is OK as before. The source is MRxSmb, event ID is 50 and status is c000020c according to XP event log. Microsoft's KB suggest bad hardware or network connection, but it's likely here. I finally managed to reproduce the problem in one of our XPs when compressing roughly 300M files in the share and store the compressed to the same one using Winrar. Compressing aborts at the middle of the process and raises the same "Delayed write failed" error. It can be reproduced only if the XP client boot after the EL-5.1 server. Same operation subsequently will be successful. After downgrading to the samba-3.0.23c-2.el5.2.0.2, it can't be reproduced anymore. No clients report similar problems so far. Version-Release number of selected component (if applicable): samba-3.0.25b-1.el5_1.4 How reproducible: Always Steps to Reproduce: 1. Store some big test files in the samba share (total size 300M in my case). Install the Winrar in XP sp2 if there isn't one. (winrar-3.61 here) 2. Reboot the EL-5.1 server. After it's ready, reboot the XP sp2 client. 3. Select all the test files in the share from XP, right click and choose compress with Winrar. Save the compressed to the same share. Actual Results: Compressing aborted and "Delayed write failed" raised. Expected Results: Compressed file should be written to the share. Additional info: smb.conf # Global parameters [global] netbios name = XXXX workgroup = XXXXXX server string = XXXXXXXXXXXXXXX log file = /var/log/samba/%m.log max log size = 500 dns proxy = No local master = Yes hosts allow = 127.0.0.1, 168.8.8.0/24 hosts deny = 0.0.0.0/0 interfaces = br0 lo bind interfaces only = Yes load printers = No show add printer wizard = No max smbd processes = 300 max connections = 300 deadtime = 15 config file = /etc/samba/smb.conf.%G passdb backend = tdbsam:/etc/samba/passdb.tdb log level = 2 username map = /etc/samba/smbusers printcap name = /dev/null disable spoolss = yes smb ports = 139 socket options = tcp_nodelay so_keepalive so_sndbuf=16384 so_rcvbuf=16384 [dep] comment = Department share path = /var/smb/%G valid users= +%G read only = No Relevent syslog: Feb 9 22:32:25 zbsd smbd[2798]: [2008/02/09 22:32:25, 0] lib/util_sock.c:write_data(562) Feb 9 22:32:25 zbsd smbd[2798]: write_data: write failure in writing to client 168.8.8.38. Error Connection reset by peer Feb 9 22:32:25 zbsd smbd[2798]: [2008/02/09 22:32:25, 0] lib/util_sock.c:send_smb(769) Feb 9 22:32:25 zbsd smbd[2798]: Error writing 4 bytes to client. -1. (Connection reset by peer) Feb 9 22:35:31 zbsd smbd[2799]: [2008/02/09 22:35:31, 0] lib/util_sock.c:read_data(534) Feb 9 22:35:31 zbsd smbd[2799]: read_data: read failure for 4 bytes to client 168.8.8.38. Error = Connection reset by peer Feb 9 22:35:31 zbsd smbd[2799]: [2008/02/09 22:35:31, 0] lib/util_sock.c:write_data(562) Feb 9 22:35:31 zbsd smbd[2799]: write_data: write failure in writing to client 168.8.8.38. Error Broken pipe Feb 9 22:35:31 zbsd smbd[2799]: [2008/02/09 22:35:31, 0] lib/util_sock.c:send_smb(769) Feb 9 22:35:31 zbsd smbd[2799]: Error writing 75 bytes to client. -1. (Broken pipe) Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:get_peer_addr(1232) Feb 9 22:35:31 zbsd smbd[2805]: getpeername failed. Error was Transport endpoint is not connected Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:get_peer_addr(1232) Feb 9 22:35:31 zbsd smbd[2805]: getpeername failed. Error was Transport endpoint is not connected Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/access.c:check_access(327) Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:get_peer_addr(1232) Feb 9 22:35:31 zbsd smbd[2805]: getpeername failed. Error was Transport endpoint is not connected Feb 9 22:35:31 zbsd smbd[2805]: Denied connection from (0.0.0.0) Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:get_peer_addr(1232) Feb 9 22:35:31 zbsd smbd[2805]: getpeername failed. Error was Transport endpoint is not connected Feb 9 22:35:31 zbsd smbd[2805]: Connection denied from 0.0.0.0 Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:write_data(562) Feb 9 22:35:31 zbsd smbd[2805]: write_data: write failure in writing to client 168.8.8.38. Error Connection reset by peer Feb 9 22:35:31 zbsd smbd[2805]: [2008/02/09 22:35:31, 0] lib/util_sock.c:send_smb(769) Feb 9 22:35:31 zbsd smbd[2805]: Error writing 5 bytes to client. -1. (Connection reset by peer
the same happened for us. we open big adobe indesign files and it happend daily. the strange thing that it's only happened for two clients. any tip or progress other then downgrade?
here is the solution which should have to included in rhel's samba too: http://marc.info/?l=samba&m=118168584629816&w=2
and the samba bug entry: https://bugzilla.samba.org/show_bug.cgi?id=4689
The above patch was part of samba 3.0.25b and is included in RHEL5 packages, I suspect this bug might be a duplicate of #435316, any chance you can try out the beta packages out soon now in the beta channel ?
(In reply to comment #4) > The above patch was part of samba 3.0.25b and is included in RHEL5 packages, > I suspect this bug might be a duplicate of #435316, any chance you can try out > the beta packages out soon now in the beta channel ? Sure, I can try it. Where can I download it?
I have some test packages I built in my people.redhat.com page, beta packages will be released through the relative RHN channel.
i read this too late, but i recompile the fedora/updates/8/SRPMS/samba-3.0.28-0.fc8.src.rpm on our server and the problem disappear so the problem somewhere between this two version. but if your packages on people will be release as a updates to rhel i'd rather test it. so i download it and install it on our server. if i don't complain in a few days than it's working:-)
it seems to working.
Over the recent months, as I've upgraded fileservers to rhel5, I've been having this problem. I tried the beta and instead of having lost data and the "Delayed" error message, I've been having lockups. I'm going to try the older version instead of the beta and will report back.
on rhel5, only rolling back to samba-3.0.23c-2.el5.2.0.2 got our production server stable. samba-3.0.25b-1.el5_1.4 caused "delayed write errors" and 3.0.28 from the beta channel has other problems. I get the hunch that it's hanging the windows xp system (sp2) on file open for write. If the beta version is about to be released, DON'T. At least for me, it causes problems which are just as serious, but are more disruptive and happen more often. If you can get 3.0.28a into rawhide I'll risk our production systems (two) and test it. We have a third fileserver running it with fedora 8 and though it's usage is quite different, it's doesn't appear to have the problems. In a year we won't need samba but until then it's critical for us.
The 3.0.28 package in the beta channel shoulkd have all the critical fixes that are in 3.0.28a. If Fedora 8 works for you (3.0.28a) I'd be *very* interested in getting more info on the problem you experienced with RHEL 5.2 beta packages. So far we got only good reports. I might have a fix in the pipeline that I could rush to an async update if it turns out it is indeed a serious problem. Can you please open a new bug and attach configuration and log files so that I can analyze the problem you are having with the beta packages?
We think it is fixed in the current available version. Also there is no reply from the reporter.