Description of problem: With current development kernels I see the following: # showmount -e smok Export list for smok: / 192.168.xx.yy/255.255.255.0 /home 192.168.xx.yy/255.255.255.0 This is expected; but # mount -v -t nfs smok:/home /mnt/srv/ mount to NFS server 'smok' failed: server is down. RPC Error: 0 ( Success ) No error messages in logs I can find. With the same version of a devel kernel but on i686 the same file system is mounted over NFS without any problems. Ditto for x86_64 but running updated FC3. For autofs results are even stranger. Corresponding directories do show up under /net/smok/ but if you will try access them they disappear. Version-Release number of selected component (if applicable): At least kernel-2.6.10-1.1154_FC4 and kernel-2.6.10-1.1149_FC4. I cannot really tell when it started.
I did a bit of tracing. Here is what strace shows in a test x86 installation (only that fragment which seems to be relevant): ..... munmap(0xb7d0a000, 4096) = 0 time(NULL) = 1109622812 socket(PF_INET, SOCK_STREAM, IPPROTO_TCP) = 3 bind(3, {sa_family=AF_INET, sin_port=htons(629), sin_addr=inet_addr("0.0.0.0")}, 16) = -1 EADDRINUSE (Address already in use) bind(3, {sa_family=AF_INET, sin_port=htons(630), sin_addr=inet_addr("0.0.0.0")}, 16) = 0 connect(3, {sa_family=AF_INET, sin_port=htons(111), sin_addr=inet_addr("192.168.xx.yy")}, 16) = 0 gettimeofday({1109622812, 897687}, NULL) = 0 write(3, "\200\0\0(}y\345\5\0\0\0\0\0\0\0\2\0\1\206\240\0\0\0\2\0"..., 44) = 44 poll([{fd=3, events=POLLIN, revents=POLLIN}], 1, 60000) = 1 read(3, "\0\0\1\214}y\345\5\0\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 500) = 456 close(3) = 0 socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP) = 3 ..... and we are off to the races with an NFS-mounte file system. And here is something from a similar system but x86_64: .... munmap(0x2aaaaae44000, 4096) = 0 socket(PF_INET, SOCK_STREAM, IPPROTO_TCP) = 3 bind(3, {sa_family=AF_INET, sin_port=htons(0), sin_addr=inet_addr("0.0.0.0")}, 16) = 0 connect(3, {sa_family=AF_INET, sin_port=htons(111), sin_addr=inet_addr("192.168.xx.yy")}, 16) = 0 write(3, "\200\0\0008nBp\205\0\0\0\0\0\0\0\2\0\1\206\240\0\0\0\2"..., 60) = 60 poll([{fd=3, events=POLLIN, revents=POLLIN}], 1, 20000) = 1 read(3, "\200\0\0\34nBp\205\0\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 500) = 32 close(3) = 0 write(2, "mount to NFS server \'smok\' faile"..., 33mount to NFS server 'smok' failed) = 33 write(2, ": server is down.\n", 18: server is down. ) = 18 .... I am not sure where the fault really lies but it sure looks different. This used to work a while ago with the same general configuration.
This is really utils-linux issue and https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152956 is a duplicate. Adding '-o udp' to mount options makes NFS work again even with those servers which fail to respond to TCP attempts.
*** This bug has been marked as a duplicate of 149790 ***