Debian Codegraph @

upstream/opensuse-tumbleweed/man8/nfsdcltrack.8

raw
1 .ie \nF \{\
2 . de IX
3 . tm Index:\\$1\t\\n%\t"\\$2"
4 ..
5 . nr % 0
6 . rr F
7 .\}
8 .el \{\
9 . de IX
10 ..
11 .\}
12 .IX Title "NFSDCLTRACK 8"
13 .TH NFSDCLTRACK 8 "2012-10-24" "" ""
14 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
15 .\" way too many mistakes in technical documents.
16 .if n .ad l
17 .nh
18 .SH "NAME"
19 nfsdcltrack \- NFSv4 Client Tracking Callout Program
20 .SH "SYNOPSIS"
21 .IX Header "SYNOPSIS"
22 nfsdcltrack [\-d] [\-f] [\-s stable storage dir] <command> <args...>
23 .SH "DESCRIPTION"
24 .IX Header "DESCRIPTION"
25 nfsdcltrack is the NFSv4 client tracking callout program. It is not necessary
26 to install this program on machines that are not acting as NFSv4 servers.
27 .PP
28 When a network partition is combined with a server reboot, there are
29 edge conditions that can cause the server to grant lock reclaims when
30 other clients have taken conflicting locks in the interim. A more detailed
31 explanation of this issue is described in \s-1RFC\s0 3530, section 8.6.3
32 and in \s-1RFC\s0 5661, section 8.4.3.
33 .PP
34 In order to prevent these problems, the server must track a small amount
35 of per-client information on stable storage. This program provides the
36 userspace piece of that functionality. When the kernel needs to manipulate
37 the database that stores this info, it will execute this program to handle
38 it.
39 .SH "OPTIONS"
40 .IX Header "OPTIONS"
41 .IP "\fB\-d\fR, \fB\-\-debug\fR" 4
42 .IX Item "-d, --debug"
43 Enable debug level logging.
44 .IP "\fB\-f\fR, \fB\-\-foreground\fR" 4
45 .IX Item "-f, --foreground"
46 Log to stderr instead of syslog.
47 .IP "\fB\-s\fR \fIstoragedir\fR, \fB\-\-storagedir\fR=\fIstorage_dir\fR" 4
48 .IX Item "-s storagedir, --storagedir=storage_dir"
49 Directory where stable storage information should be kept. The default
50 value is \fI/var/lib/nfs/nfsdcltrack\fR.
51 .SH "COMMANDS"
52 .IX Header "COMMANDS"
53 nfsdcltrack requires a command for each invocation. Supported commands
54 are:
55 .IP "\fBinit\fR" 4
56 .IX Item "init"
57 Initialize the database. This command requires no argument.
58 .IP "\fBcreate\fR" 4
59 .IX Item "create"
60 Create a new client record (or update the timestamp on an existing one). This command requires a hex-encoded nfs_client_id4 as an argument.
61 .IP "\fBremove\fR" 4
62 .IX Item "remove"
63 Remove a client record from the database. This command requires a hex-encoded nfs_client_id4 as an argument.
64 .IP "\fBcheck\fR" 4
65 .IX Item "check"
66 Check to see if a nfs_client_id4 is allowed to reclaim. This command requires a hex-encoded nfs_client_id4 as an argument.
67 .IP "\fBgracedone\fR" 4
68 .IX Item "gracedone"
69 Remove any unreclaimed client records from the database. This command requires a epoch boot time as an argument.
70 .SH "EXTERNAL CONFIGURATION"
71 The directory for stable storage information can be set via the file
72 .B /etc/nfs.conf
73 by setting the
74 .B storagedir
75 value in the
76 .B nfsdcltrack
77 section. For example:
78 .in +5
79 [nfsdcltrack]
80 .br
81 storagedir = /shared/nfs/nfsdcltrack
82 .in -5
83 Debugging to syslog can also be enabled by setting "debug = 1" in this file.
84 .SH "LEGACY TRANSITION MECHANISM"
85 .IX Header "LEGACY TRANSITION MECHANISM"
86 The Linux kernel NFSv4 server has historically tracked this information
87 on stable storage by manipulating information on the filesystem
88 directly, in the directory to which \fI/proc/fs/nfsd/nfsv4recoverydir\fR
89 points. If the kernel passes the correct information, then nfsdcltrack
90 can use it to allow a seamless transition from the old client tracking
91 scheme to the new one.
92 .PP
93 On a \fBcheck\fR operation, if there is no record of the client in the
94 database, nfsdcltrack will look to see if the \fB\s-1NFSDCLTRACK_LEGACY_RECDIR\s0\fR
95 environment variable is set. If it is, then it will fetch that value and
96 see if a directory exists by that name. If it does, then the check
97 operation will succeed and the directory will be removed.
98 .PP
99 On a \fBgracedone\fR operation, nfsdcltrack will look to see if the
100 \&\fB\s-1NFSDCLTRACK_LEGACY_TOPDIR\s0\fR environment variable is set. If it is, then
101 it will attempt to clean out that directory prior to exiting.
102 .PP
103 Note that this transition is one-way. If the machine subsequently reboots
104 back into an older kernel that does not support the nfsdcltrack upcall
105 then the clients will not be able to recover their state.
106 .SH "NOTES"
107 .IX Header "NOTES"
108 This program requires a kernel that supports the nfsdcltrack usermodehelper
109 upcall. This support was first added to mainline kernels in 3.8.
110 .SH "AUTHORS"
111 .IX Header "AUTHORS"
112 nfsdcltrack was developed by Jeff Layton <[email protected]>.