1gfs_controld(8) System Manager's Manual gfs_controld(8)
2
6 gfs_controld - daemon that manages mounting, unmounting, recovery and
7 posix locks
8
11 gfs_controld [OPTION]...
12
15 GFS lives in the kernel, and the cluster infrastructure (cluster mem‐
16 bership and group management) lives in user space. GFS in the kernel
17 needs to adjust/recover for certain cluster events. It's the job of
18 gfs_controld to receive these events and reconfigure gfs as needed.
19 gfs_controld controls and configures gfs through sysfs files that are
20 considered gfs-internal interfaces; not a general API/ABI.
21 Mounting, unmounting and node failure are the main cluster events that
23 gfs_controld controls. It also manages the assignment of journals to
24 different nodes. The mount.gfs and umount.gfs programs communicate
25 with gfs_controld to join/leave the mount group and receive the neces‐
26 sary options for the kernel mount.
27 GFS also sends all posix lock operations to gfs_controld for process‐
29 ing. gfs_controld manages cluster-wide posix locks for gfs and passes
30 results back to gfs in the kernel.
31
34 Optional cluster.conf settings are placed in the <gfs_controld> sec‐
35 tion.
36 Posix locks
39 Heavy use of plocks can result in high network load. The rate at which
40 plocks are processed are limited by the plock_rate_limit setting, which
41 limits the maximum plock performance, and limits potentially excessive
42 network load. This value is the maximum number of plock operations a
43 single node will process every second. To achieve maximum posix lock‐
44 ing performance, the rate limiting should be disabled by setting it to
45 0. The default value is 100.
46 <gfs_controld plock_rate_limit="100"/>
48 To optimize performance for repeated locking of the same locks by pro‐
50 cesses on a single node, plock_ownership can be set to 1. The default
51 is 0. If this is enabled, gfs_controld cannot interoperate with older
52 versions that did not support this option.
53 <gfs_controld plock_ownership="1"/>
55 Three options can be used to tune the behavior of the plock_ownership
57 optimization. All three relate to the caching of lock ownership state.
58 Specifically, they define how aggressively cached ownership state is
59 dropped. More caching of ownership state can result in better perfor‐
60 mance, at the expense of more memory usage.
61 drop_resources_time is the frequency of drop attempts in milliseconds.
63 Default 10000 (10 sec).
64 drop_resources_count is the maximum number of items to drop from the
66 cache each time. Default 10.
67 drop_resources_age is the time in milliseconds a cached item should be
69 unused before being considered for dropping. Default 10000 (10 sec).
70 <gfs_controld drop_resources_time="10000" drop_resources_count="10"
72 drop_resources_age="10000"/>
73
77 -D Run the daemon in the foreground and print debug statements to
78 stdout.
79 -P Enable posix lock debugging messages.
81 -w Disable the "withdraw" feature.
83 -p Disable posix lock handling.
85 -l <num>
87 Limit the rate at which posix lock messages are sent to <num>
88 messages per second. 0 disables the limit and results in the
89 maximum performance of posix locks. Default 100.
90 -o <num>
92 Enable (1) or disable (0) plock ownership optimization. Default
93 0. All nodes must run with the same value.
94 -t <ms>
96 Ownership cache tuning, drop resources time (milliseconds).
97 Default 10000.
98 -c <ms>
100 Ownership cache tuning, drop resources count. Default 10.
101 -a <ms>
103 Ownership cache tuning, drop resources age (milliseconds).
104 Default 10000.
105 -h Print out a help message describing available options, then
107 exit.
108 -V Print the version information and exit.
110
113 The gfs_controld daemon keeps a circular buffer of debug messages that
114 can be dumped with the 'group_tool dump gfs' command.
115 The state of all gfs posix locks can also be dumped from gfs_controld
117 with the 'group_tool dump plocks <fsname>' command.
118
121 groupd(8), group_tool(8)
122 gfs_controld(8)