Red Hat Bugzilla – Bug 197563
Samba over nfs troubles, hangs Windows clients
Last modified: 2008-01-15 06:41:01 EST
Description of problem: Samba over nfs troubles, it hangs Windows clients as soon as I try to open files via Samba which in turn reside on NFS. I have tried the same on FC5 and there it works. Version-Release number of selected component (if applicable): nfs-utils-1.0.6-65.EL4 samba-3.0.10-1.4E.6 2.6.9-34.0.1.ELsmp How reproducible: always Steps to Reproduce: 1. NFS export: /srv/nfsexp 127.0.0.1(rw) 2. mount it: 127.0.0.1:/srv/nfsexp /srv/samba/nfs nfs rw,sync 0 0 3. make a samba share and mount it via Windows: smb.conf: ... [nfstest] path = /srv/samba writable = yes valid users = tr Actual results: As soon as I start to read and write files below /srv/samba/nfs via Samba on Windows the connection hangs, I get read timeout in Explorer Expected results: Read and write fine Additional info: NFS is functional, I can read/write in /srv/samba/nfs as user tr and see the changes in /srv/nfsexp. Samba is functional, I can read/write with windows in /srv/samba but not in /srv/samba/nfs. I played with variuos nfs export options as well as samba locking options, always the same problem.
An additional note: this is just a simple test case, reproducing the problem with just one host involved to keep the configuration as simple as possible. The same happens if NFS is mounted on a different host and the samba share is exported there.
In regards to comment #1, in order to reproduce this, which host needs to be RHEL4? Does this problem reproduce when the NFS server is RHEL4, the NFS client is RHEL4, or only when both are on RHEL4? You mentioned that it worked with FC5, which piece did you replace with a FC5 machine when you discovered this?
Just hit this same problem. It happens only when NFSv4 mounts are active. Just switched from NFSv3 to NFSv4 mounts and now an application that reads a huge file over a Samba share locks up on read requests. Sometimes stop/start cycling Samba helps but the most reliable fix is to put the NFS shares back to NFSv3 and reboot. Here are some messages related to the hang. Sep 26 11:14:24 keira smbd[18850]: [2007/09/26 11:14:24, 0] lib/util_sock.c:get_peer_addr(1000) Sep 26 11:14:24 keira smbd[18850]: getpeername failed. Error was Transport endpoint is not connected Sep 26 11:14:24 keira smbd[18850]: [2007/09/26 11:14:24, 0] lib/util_sock.c:write_socket_data(430) Sep 26 11:14:24 keira smbd[18850]: write_socket_data: write failure. Error = Connection reset by peer Sep 26 11:14:24 keira smbd[18850]: [2007/09/26 11:14:24, 0] lib/util_sock.c:write_socket(455) Sep 26 11:14:24 keira smbd[18850]: write_socket: Error writing 4 bytes to socket 5: ERRNO = Connection reset by peer Sep 26 11:14:24 keira smbd[18850]: [2007/09/26 11:14:24, 0] lib/util_sock.c:send_smb(647) Sep 26 11:14:24 keira smbd[18850]: Error writing 4 bytes to client. -1. (Connection reset by peer)
Wait! I apologize as I attributed the incorrect cause to the effect. Now I see the problem happens with the NFSv3 mounts active, or with *no* NFS mounts active. The more I dig the stranger this gets, so it's probably nothing to do with this issue. Sorry.
No response from the original reporter for several months. Closing with resolution of INSUFFICIENT_DATA. Please reopen if you're still able to reproduce this and can provide the info requested.