[Samba] Random share disconnects
mlondon at alltec.com
Wed Jan 11 19:12:00 GMT 2006
We've got a site running a windows domain with a samba PDC and we're seeing a bunch of random disconnects from shares on Windows XP Pro workstations.
The server config is:
SuSE Enterprise Linux 9 (SLES9) with:
samba-3.0.14a-0.4 (using ldap)
openldap2-2.2.24-4.5 (using a bdb backend)
There's also a heartbeat setup which will shutdown smbd and nmbd on one machine and bring up a backup machine, but that shouldn't affect anything, as logs show it's not doing anything.
Client machines are mostly running Windows XP Pro, but there are a handful of Win98 machines around as well.
The problem has only been reported by WinXP users and it appears to occur to random users at random times.
Users will be working at their PC when all their network mapped drives go offline. If they open up the drive in explorer, they get a blank window. In most cases, they also get a bubble saying "You are now working offline" (they have My Documents on a mapped drive, which automatically seems to enable offline files for that folder).
Sometimes the shares come back on their own; sometimes the user has to reboot.
Looking at the logs on the server, there doesn't appear to be anything odd. There's a whole bunch of connection reset by peer notices, e.g.
[2006/01/11 18:22:11, 0] lib/util_sock.c:write_socket_data(430)
[2006/01/11 18:22:11, 2] smbd/sesssetup.c:setup_new_vc_session(608)
setup_new_vc_session: New VC == 0, if NT4.x compatible we would close all old
write_socket_data: write failure. Error = Connection reset by peer
[2006/01/11 18:22:11, 0] lib/util_sock.c:write_socket(455)
[2006/01/11 18:22:11, 2] lib/module.c:do_smb_load_module(66)
Module '/usr/lib/samba/vfs/recycle.so' loaded
write_socket: Error writing 4 bytes to socket 45: ERRNO = Connection reset by
[2006/01/11 18:22:11, 0] lib/util_sock.c:send_smb(647)
Error writing 4 bytes to client. -1. (Connection reset by peer)
[2006/01/11 18:22:11, 2] smbd/server.c:exit_server(609)
But I don't know if this co-relates to shares going offline, because the users can't give me the times when they saw problems.
This doesn't seem to affect every user, it doesn't seem to co-relate to heavy system load or heavy network utilization. The network interface doesn't show any errors and the site where this is most prevalent is a fully switched network with the everyone (including the server) on a single switch stack. We were seeing this problem at their old site, where the network contained a lot of small switches and hubs, but the problem is still present at the new site.
Any thoughts on how to troubleshoot or fix the problem would be most appreciated.
More information about the samba