Created attachment 353556 [details] upgraded packages + emerge --info I have had net-fs/nfs-utils-1.2.6 installed for over six months. I last successfully started it in May 2013. This start generated the following syslog output: 2013 May 16 22:42:42 daemon: Version 1.2.6 starting 2013 May 16 22:42:42 daemon: Flags: TI-RPC 2013 May 16 22:42:42 daemon: Running as root. chown /var/lib/nfs to choose different user 2013 May 16 22:42:42 kern: [ 946.537481] RPC: Registered named UNIX socket transport module. 2013 May 16 22:42:42 kern: [ 946.537491] RPC: Registered udp transport module. 2013 May 16 22:42:42 kern: [ 946.537495] RPC: Registered tcp transport module. 2013 May 16 22:42:42 kern: [ 946.537498] RPC: Registered tcp NFSv4.1 backchannel transport module. 2013 May 16 22:42:43 kern: [ 947.016586] Installing knfsd (copyright (C) 1996 okir@monad.swb.de). 2013 May 16 22:42:43 daemon: nfsdreopen: Opening '/proc/net/rpc/nfs4.nametoid/channel' failed: errno 2 (No such file or directory) 2013 May 16 22:42:43 daemon: nfsdreopen: Opening '/proc/net/rpc/nfs4.idtoname/channel' failed: errno 2 (No such file or directory) 2013 May 16 22:42:43 daemon: Version 1.2.6 starting 2013 May 16 22:42:43 daemon: Version 1.2.6 starting 2013 May 16 22:45:16 kern: [ 1099.767040] svc: 192.168.1.3, port=48556: unknown version (3 for prog 100003, nfsd) 2013 May 16 22:45:16 kern: [ 1099.770485] svc: 192.168.1.3, port=50897: unknown version (3 for prog 100003, nfsd) 2013 May 16 22:45:16 daemon: authenticated mount request from 192.168.1.3:732 for /home (/home) Since that time, I have installed or upgraded 296 packages. nfs-utils was not one of the 296 packages; it remained at version 1.2.6, and I did not upgrade the kernel. Since those upgrades, NFS now fails to start: # date; /etc/init.d/nfs start Wed Jul 17 20:52:19 CDT 2013 * Starting rpcbind ... [ ok ] * Starting NFS statd ... [ ok ] * Setting up RPC pipefs ... [ ok ] * Starting idmapd ... [ ok ] * Mounting nfsd filesystem in /proc ... [ ok ] * Exporting NFS directories ... exportfs: /etc/exports [1]: Neither 'subtree_check' or 'no_subtree_check' specified for export "192.168.1.0/255.255.255.0:/home". Assuming default behaviour ('no_subtree_check'). NOTE: this default has changed since nfs-utils version 1.0.x [ ok ] * Starting NFS mountd ... [ ok ] * Starting NFS daemon ... [ !! ] * Starting NFS smnotify ... [ ok ] * ERROR: nfs failed to start This attempt produced the following syslog output: 2013 Jul 17 20:52:19 daemon: Version 1.2.6 starting 2013 Jul 17 20:52:19 daemon: Flags: TI-RPC 2013 Jul 17 20:52:19 daemon: Running as root. chown /var/lib/nfs to choose different user 2013 Jul 17 20:52:20 kern: [1643866.317453] RPC: Registered named UNIX socket transport module. 2013 Jul 17 20:52:20 kern: [1643866.317462] RPC: Registered udp transport module. 2013 Jul 17 20:52:20 kern: [1643866.317466] RPC: Registered tcp transport module. 2013 Jul 17 20:52:20 kern: [1643866.317470] RPC: Registered tcp NFSv4.1 backchannel transport module. 2013 Jul 17 20:52:21 kern: [1643867.171293] Installing knfsd (copyright (C) 1996 okir@monad.swb.de). 2013 Jul 17 20:52:21 daemon: nfsdreopen: Opening '/proc/net/rpc/nfs4.nametoid/channel' failed: errno 2 (No such file or directory) 2013 Jul 17 20:52:21 daemon: nfsdreopen: Opening '/proc/net/rpc/nfs4.idtoname/channel' failed: errno 2 (No such file or directory) 2013 Jul 17 20:52:21 daemon: Version 1.2.6 starting 2013 Jul 17 20:52:21 kern: [1643867.618222] lockd_up: svc_rqst allocation failed, error=-12 2013 Jul 17 20:52:21 kern: [1643867.619401] nfsd: last server has exited, flushing export cache 2013 Jul 17 20:52:21 daemon: error starting threads: errno 12 (Cannot allocate memory) 2013 Jul 17 20:52:21 daemon: Version 1.2.6 starting 2013 Jul 17 20:52:21 daemon: ERROR: nfs failed to start The successful start generated the nfsdreopen errors as well, so that doesn't seem to be related to the failure. "Cannot allocate memory" seems the most relevant message, but there is about 3G of physical+virtual memory available. (The machine has only 1.2G of physical memory, but it had the same limitation last time NFS started as well.) I have not changed anything about the NFS configuration on the system. /etc/exports has a 2012 timestamp. /etc/conf.d/nfs was modified May 17 01:35, but this must have been done by the NFS daemon itself, as I did not touch this file. So it seems one of the other package upgrades has caused the failure, but I don't know how to narrow it any further. I am attaching a list of these packages. There may not be enough information here to debug the problem; please let me know what else I can provide that will help.
Please post your `emerge -vp net-fs/nfs-utils' output in a comment.
Hmm, I'm not sure why the summary was changed to highlight the nfsdreopen error; as the bug description shows, this error occurs in both the successful and unsuccessful NFS starts, so it does not seem to be the source of the start-up failure. The lockd_up error is unique to the failure case. (In reply to Jeroen Roovers from comment #1) # emerge --color n -vp net-fs/nfs-utils These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R ] net-fs/nfs-utils-1.2.6 USE="ipv6 nfsv4 tcpd -caps -kerberos -nfsdcld -nfsidmap -nfsv41 (-selinux)" 0 kB Total: 1 package (1 reinstall), Size of downloads: 0 kB #
Hi, what further information can I provide that will help in debugging this issue?
Hello, I am still unable to start NFS. My latest attempt produced output to "/etc/init.d/nfs start" and syslog entries that were both essentially the same as what I posted in comment #0. Please tell me what I need to do to help debug this problem.
(In reply to Dave Kemper from comment #2) you're building with USE=nfsv4, but the logging output indicates you didn't enable NFSv4 in your kernel. either disable it in your nfs-utils emerge, or enable support in your kernel and reboot.
SpanKY, thank you for the reply. I do have NFSv4 enabled in the kernel, but not NFSv4.1: $ fgrep NFS_V4 /usr/src/linux/.config CONFIG_NFS_V4=y # CONFIG_NFS_V4_1 is not set $ ls -l /usr/src/linux/.config -rw-r--r-- 1 root root 70903 Aug 28 2012 /usr/src/linux/.config The NFS daemon has successfully started since Aug 2012, and the kernel config has been unchanged since that time, so I'm not sure how this could be a kernel problem. Am I misunderstanding?
(In reply to Dave Kemper from comment #6) if you have CONFIG_NFSD_V4=y then you should be able to run `sudo cat /proc/net/rpc/nfs4.nametoid/channel` and have it work
(In reply to SpanKY from comment #7) > if you have CONFIG_NFSD_V4=y Aha, NFS_V4 is enabled but NFSD_V4 is not. I'll fix that. Still, the kernel has been configured the same way since August 2012, and net-fs/nfs-utils-1.2.6 has been installed and unchanged since December 2012. Yet NFS worked in May 2013 (see comment #0), so the CONFIG_NFSD_V4 setting can't be the whole story.
