samba - Sep 2002 - Read failure for 4. Error = connection reset by peer.

I am having a SERIOUS problem with our new 2.2.5 server. In fact I will 
probably be forced to move to a Windows 2000 Server platform, as the samba 
server is practically useless.  Work on any files on the share cannot be 
done without major risk of losing data. I get these two errors all day:

[2002/09/19 21:04:21, 0] lib/util_sock.c:read_data(436)
   read_data: read failure for 4. Error = Connection reset by peer

and

[2002/09/20 00:58:43, 0] lib/util_sock.c:read_data(436)
   read_data: read failure for 4. Error = Connection timed out

Most of them are the connection reset by peer. I get more than 25 a day at 
times, at intervals ranging from 5 minutes to several hours. If someone has 
a file open during that time and try to save I get a bunch of oplock log 
errors, their application crashes and all data is lost. I've searched and 
searched w/ google, no one seems to have any solutions. I've tried some 
keepalive options. Both socket level and application level, both failed. 
Somone mentioned a registry fix but that also seems to have failed. My 
clients are mostly Windows 2k with a couple of XP's.

Frustrated,
Jason Valenzuela

On Mon, Sep 23, 2002 at 06:44:35PM -0400, Jason Valenzuela
wrote:
> I am having a SERIOUS problem with our new 2.2.5 server. In fact I will 
> probably be forced to move to a Windows 2000 Server platform, as the samba 
> server is practically useless.  Work on any files on the share cannot be 
> done without major risk of losing data. I get these two errors all day:
> 
> [2002/09/19 21:04:21, 0] lib/util_sock.c:read_data(436)
>    read_data: read failure for 4. Error = Connection reset by peer
> 
> and
> 
> [2002/09/20 00:58:43, 0] lib/util_sock.c:read_data(436)
>    read_data: read failure for 4. Error = Connection timed out
You have a serious problem with your network.  There is very little
a server can do if the clients 'go away'.  Do however check that
you are not using one of the bad releases of nss_ldap, and double
check that 2.2.6pre2 doesn't fix anything.

Assuming you rule out software, start looking for bad network gear.

Andrew Bartlett

I'll offer this without asking what your setup is.

In the Global section, add:
oplocks = no
level2 oplocks = no
kernel oplocks = no

Then killall -HUP smbd
Then killall -HUP nmbd
to force Samba to reread the config file.

If you already did this then others will have to help.  This fixed that problem
every time I ran into it.


Jason Valenzuela (jason.valenzuela@worldnet.att.net)
wrote*:
>
>I am having a SERIOUS problem with our new 2.2.5 server. In fact I will
>probably be forced to move to a Windows 2000 Server platform, as the samba
>server is practically useless.  Work on any files on the share cannot be
>done without major risk of losing data. I get these two errors all day:
>
>[2002/09/19 21:04:21, 0] lib/util_sock.c:read_data(436)
>   read_data: read failure for 4. Error = Connection reset by peer
>
>and
>
>[2002/09/20 00:58:43, 0] lib/util_sock.c:read_data(436)
>   read_data: read failure for 4. Error = Connection timed out
>
>Most of them are the connection reset by peer. I get more than 25 a day at
>times, at intervals ranging from 5 minutes to several hours. If someone has
>a file open during that time and try to save I get a bunch of oplock log
>errors, their application crashes and all data is lost. I've searched
and
>searched w/ google, no one seems to have any solutions. I've tried some
>keepalive options. Both socket level and application level, both failed.
>Somone mentioned a registry fix but that also seems to have failed. My
>clients are mostly Windows 2k with a couple of XP's.
>
>Frustrated,
>Jason Valenzuela
>
>
--
Bob Crandell
Assured Computing
When you need to be sure.
Voice 541-689-9159
FAX   240-371-7237
bob@assuredcomp.com
www.assuredcomp.com
Eugene, Or. 97402

You can try setting 

deadtime=10

I think what is happening is the the Windows 2000 connection is timing out at
the same time the Samba connection does and things get confused.  Reducing the
dead time seemed to help but not completely solve my problem with the connection
reset by peer message.

Dan Smith
Sr. System Analyst
Pinnacle West Shared Services
"If you can't get your work done in 24 hours a day,
work nights"



-----Original Message-----
From: Jason Valenzuela [mailto:jason.valenzuela@worldnet.att.net]
Sent: Monday, September 23, 2002 3:45 PM
To: samba@lists.samba.org
Subject: [Samba] Read failure for 4. Error = connection reset by peer.


I am having a SERIOUS problem with our new 2.2.5 server. In fact I will 
probably be forced to move to a Windows 2000 Server platform, as the samba 
server is practically useless.  Work on any files on the share cannot be 
done without major risk of losing data. I get these two errors all day:

[2002/09/19 21:04:21, 0] lib/util_sock.c:read_data(436)
   read_data: read failure for 4. Error = Connection reset by peer

and

[2002/09/20 00:58:43, 0] lib/util_sock.c:read_data(436)
   read_data: read failure for 4. Error = Connection timed out

Most of them are the connection reset by peer. I get more than 25 a day at 
times, at intervals ranging from 5 minutes to several hours. If someone has 
a file open during that time and try to save I get a bunch of oplock log 
errors, their application crashes and all data is lost. I've searched and 
searched w/ google, no one seems to have any solutions. I've tried some 
keepalive options. Both socket level and application level, both failed. 
Somone mentioned a registry fix but that also seems to have failed. My 
clients are mostly Windows 2k with a couple of XP's.

Frustrated,
Jason Valenzuela

-- 
To unsubscribe from this list go to the following URL and read the
instructions:  http://lists.samba.org/mailman/listinfo/samba