Hi, I run Samba 2.0.5a on Linux 2.2.12 (Slackware). My clients get an error regularly (Windows95) "Network path not found" and then they lose their connection completely. The server logs the following: [1999/09/07 17:30:01, 1] smbd/service.c:make_connection(521) sbit12 (137.56.40.139) connect to service sbitdata as user sbit (uid=1000, gid=100) (pid 134) [1999/09/07 17:33:10, 0] lib/util_sock.c:read_socket_data(507) read_socket_data: recv failure for 4. Error = Connection reset by peer [1999/09/07 17:33:10, 1] smbd/service.c:close_cnum(557) sbit12 (0.0.0.0) closed connection to service sbitdata This happens unpredictable, very strange. Does anyone know how I can solve this problem, because it is very annoying :( and keeps us from working. I searched the internet, but there were only people with the same problem, but no solutions. Thanks in advance, Martin -- Martin Schapendonk, M.Schapendonk@kub.nl, Phone: +31 (0)13-4678293 Student Information Management and Technology at Tilburg University WEBMaster of the Students-WWW-server stuwww.kub.nl / PGP on request
Hi,
I am a newbie Samba administrator, so please forgive this if it is a silly
question, but some of our clients are getting "Connection reset by
peer" when
doing a large file copy from an NT 4.0 workstation to a Samba 2.0.5a server.
This happens from a cmd window ("DOS box"), and it also happens when
there is a
very compute-intensive operation going on from the PC to a networked file (ie. a
long time between responses).
I thought the "Deadtime" parameter was causing the connection resets,
but that
was set to 0, which should not cause any timeouts.
Does anyone have any ideas on this?
Also, is there a searchable archive for this mail list? I've gone to the
Samba
lists archive site, but after waiting for several archives to load (we only have
a slow Internet connection) and searching each one for this type of problem,
I've given up...
-Thanks,
-Tom Haws
_________________________________________________________________
Tom Haws
System Administrator Inform Network for Management Systems
Tel: (250) 562-2628 1579 9th Ave, Prince George, B.C. V2L 3R8
Fax: (250) 562-6942
_________________________________________________________________
I administer Samba 2.0.7 on Debian Linux (kernel 2.2.17, libc6 2.2). When trying to mount samba shares from Windows 98 workstations sometime the process failed with error message: [2000/12/26 11:24:28, 0] lib/util_sock.c:read_socket_data(477) read_socket_data: recv failure for 4. Error = Connection reset by peer I've looked throwgh the samba, samba-binaries, and samba-technical mailing lists since 1999 and found a lot of questions on the subject but no answer, which could help to solve the problem. As samba guru keeps silence my question is to thouse who wrote to the lists earlier and know now how to solve the problem. Help, please. Thank you, Mikhail.
hallo, Mikhail,
Telling you WHAT the message means is no problem - Samba had a connection
open to the pc, and had hung a read on the socket (file descriptor
representing the connection) and was expecting information back; what samba
found was an error return when it tried to read from the socket, indicating
that the connection had been reset FROM the client side. That is to say,
last thing samba knew, the connection was good, it send something to the
client, and when it tried to read back reply from the client, the connection
was gone. WHY? that's of course the real question, and for that I have no
answer. A couple of questions that might help others on this list make some
guesses:
1. when this happens, WHAT does the client have to do to recover?
- can the client (win98) just try to access (mount) the shares again, and
it works?
- or does the client have to reboot, or what?
2. when this happens do ALL the workstations have the same problem, or is is
just one or
two, while the rest of the clients are still fine?
3. Is there any indication in the system logs, or your network statistics,
that there are any
excessive tcpip retries or errors around the time of the failure, or any
other resource
shortages? I'm not familiar with Debian Linux, but perhaps someone on
the list could
point you to the utilities Debian supplies to check this out.
Hope this helps,
Don
-----Original Message-----
From: Romanenko M.A. [mailto:mikhail@angg.ru]
Sent: Tuesday, December 26, 2000 12:50 AM
To: samba@us5.samba.org
Subject: Connection reset by peer
I administer Samba 2.0.7 on Debian Linux (kernel 2.2.17, libc6 2.2). When
trying to mount samba shares from Windows 98 workstations sometime the
process failed with error message:
[2000/12/26 11:24:28, 0] lib/util_sock.c:read_socket_data(477)
read_socket_data: recv failure for 4. Error = Connection reset by peer
I've looked throwgh the samba, samba-binaries, and samba-technical mailing
lists since 1999 and found a lot of questions on the subject but no answer,
which could help to solve the problem. As samba guru keeps silence my
question is to thouse who wrote to the lists earlier and know now how to
solve the problem.
Help, please.
Thank you, Mikhail.
<!-- newbie alert, please be patient ;) -->
Hi,
Having a problem with samba 2.2.1a on Suse 7.3. Things all appear to be
working fine, except for users who use a largeish MSAccess database, and
have it open on their machines all the time. After an indeterminate period
of time, (5-20 mins) they are complaining that it crashes with a
file/network access error.
I tried restarting samba while someone was connected, and it had the same
result so it looks like samba is kicking them off after x period of time.
I was getting a lot of errors re: oplocks so I switched these off, however
I think this was a symptom, not a cause. The two main errors I'm getting
now, are 'could not fetch trust account password for domain x' and
'connection reset by peer'. I'm not sure if the former error has
anything
to do with the latter (see long time gap in following log). The last 4
lines of the samba log for one of the affected machines are as follows:
[2001/11/15 14:50:03, 0] smbd/password.c:domain_client_validate(1563)
domain_client_validate: could not fetch trust account password for
domain WIDGITSOFTWARE
[2001/11/15 14:58:16, 0] smbd/password.c:domain_client_validate(1563)
domain_client_validate: could not fetch trust account password for
domain WIDGITSOFTWARE
[2001/11/15 14:58:42, 0] lib/util_sock.c:read_socket_data(478)
read_socket_data: recv failure for 4. Error = Connection reset by peer
[2001/11/15 15:10:04, 0] lib/util_sock.c:read_socket_data(478)
read_socket_data: recv failure for 4. Error = Connection reset by peer
This appears to be happening on all machines that have these persistent
connections to the database. Everyone else who uses normal file-sharing
isn't getting the 'connection reset' errors.
Any ideas what could be causing this? I'm scratching my head on this one.
Thanks,
Simon
smb.conf: (comments removed)
[global]
workgroup = WIDGITSOFTWARE
security = domain
domain logons = yes
domain master = yes
encrypt passwords = yes
os level = 2
kernel oplocks = No
level2 oplocks = No
oplocks = No
guest account = Nobody
map to guest = Bad User
create mask = 0770
directory mask = 0770
hosts allow = 192.168.1.
netbios aliases = FILESERVER
time server = yes
deadtime = 0
unix password sync = yes
logon path = \\%L\profiles\%U
logon home = \\%L\%U
log file = /var/log/samba/%m
printing = LPRNG
printcap name = /etc/printcap
load printers = Yes
socket options = SO_KEEPALIVE IPTOS_LOWDELAY TCP_NODELAY
interfaces = 192.168.1.1/255.255.255.0
wins support = yes
[netlogon]
comment = Network Logon Service
path = /var/lib/samba/netlogon
[profiles]
comment = Network Profiles Service
path = /var/lib/samba/profiles
browseable = Yes
writeable = Yes
character set = ISO8859-15
client code page = 850
[homes]
comment = Home Directories
read only = No
create mask = 0640
directory mask = 0750
browseable = Yes
writeable = Yes
[admin]
comment = Administration Stuff
read only = No
browseable = Yes
writeable = Yes
path=/export/raid/adminshare
[development]
comment = Development Stuff
read only = No
browseable = Yes
writeable = Yes
path=/export/raid/devshare
[sourcesafe]
comment = Microsoft Visual Sourcesafe share
read only = no
browseable = Yes
writeable = Yes
path=/export/raid/sourcesafe
---------------------------------------------
Simon Detheridge
R&D Programmer
Widgit Software Ltd. <http://www.widgit.com/>