1CONFDB_KEYS(8) Corosync Cluster Engine Programmer's Manual CONFDB_KEYS(8)
2
6 confdb_keys - Overview of keys stored in the Configuration Database
7
10 There are roughly 3 types of keys stored in ConfDB:
11 * Values stored in config file.
13 * Runtime statistics.
15 * Other user created values.
17
20 internal_configuration.*
21 Internal configuration data. It's only useful for getting list
22 of loaded services.
23 logging.*
26 Values read from configuration file. It's possible to change
27 them at runtime. If subsystem specific configuration is needed,
28 each subsys must have one object with subsys key. Other keys are
29 same as in config file.
30 runtime.blackbox.*
34 Trigger keys for store fplay data. It's recommended to use
35 corosync-blackbox command to change keys in this object.
36 runtime.connections.*
39 There are informations about total number of active connections
40 in given moment in active key, number of closed connections dur‐
41 ing whole runtime of corosync in closed key and informations
42 about each active IPC connection.
43 runtime.connections.ID.*
46 Each IPC connection has unique ID. This is in form
47 [[short_name:][PID:]internal_id. On some platforms, short_name
48 and PID are not filled and only internal_id is used.
49 Typical keys in object are:
51 client_pid containing PID of IPC connection (unavailable on some
53 platforms).
54 dispatched with number of dispatched messages.
56 invalid_request is number of requests made by IPC which are
58 invalid (calling non-existing call, ...).
59 name containing short name of IPC connection (unavailable on
61 some platforms).
62 overload is number of requests which were not processed because
64 of overload.
65 queue_size contains number of messages in queue waiting for
67 send.
68 recv_retries is total number of interrupted receives.
70 requests contains number of requests made by IPC.
72 responses is number of responses sent to IPC client.
74 send_retries contains total number of interrupted sends.
76 service_id contains ID of service which IPC is connected to.
78 runtime.services.*
81 Objects with statistics for service engines. Each service has
82 it's own subobject with name SERVICE, where SERVICE is lower
83 case name of service. Inside object is number of received and
84 send messages by corosync engine in format runtime.services.SER‐
85 VICE.EXEC_CALL.rx and runtime.services.SERVICE.EXEC_CALL.tx,
86 where EXEC_CALL is internal id of service call (so for example 3
87 in cpg service is receive of multicast message from other
88 nodes).
89 runtime.totem.pg.mrp.srp.*
92 Object with statistics about totem. Typical keys:
93 commit_entered Number of times processor entered COMMIT state.
95 commit_token_lost Number of times processor lost token in COMMIT
97 state.
98 consensus_timeouts How many times processor timeouted making
100 consensus about membership.
101 continuous_gather How many times was processor not able to reach
103 consensus.
104 firewall_enabled_or_nic_failure Set to 1 when processor was not
106 able to reach consensus for long time. Usual reason is badly
107 configured firewall or connection failure.
108 gather_entered Number of times processor entered GATHER state.
110 gather_token_lost Number of times processor lost token in GATHER
112 state.
113 mcast_retx Number of retransmitted messages.
115 mcast_rx Number of received multicast messages.
117 mcast_tx Number of transmitted multicast messages.
119 memb_commit_token_rx Number of received commit tokens.
121 memb_commit_token_tx Number of transmitted commit tokens.
123 memb_join_rx Number of received join messages.
125 memb_join_tx Number of transmitted join messages.
127 memb_merge_detect_rx Number of received member merge messages.
129 memb_merge_detect_tx Number of transmitted member merge mes‐
131 sages.
132 orf_token_rx Number of received orf tokens.
134 orf_token_tx Number of transmitted orf tokens.
136 recovery_entered Number of times processor entered recovery.
138 recovery_token_lost Number of times token was lost in recovery
140 state.
141 rx_msg_dropped Number of received messages which was dropped
143 because they were not expected (as example multicast message in
144 commit state).
145 token_hold_cancel_rx Number of received token hold cancel mes‐
147 sages.
148 token_hold_cancel_tx Number of transmitted token hold cancel
150 messages.
151 mtt_rx_token Mean transit time of token in milliseconds. In
153 other words, time between two consecutive token receives.
154 avg_token_workload Average time in milliseconds of holding time
156 of token on current processor.
157 avg_backlog_calc Average number of not yet sent messages of cur‐
159 rent processor.
160 runtime.totem.pg.mrp.srp.members.*
163 Object containing members of totem single ring protocol. Each
164 member key has format runtime.totem.pg.mrp.srp.mem‐
165 bers.NODEID.KEY, where key is one of:
166 ip IP address of member. It's stored in format r(RING_ID)
168 ip(IP_ADDRESS).
169 join_count Number of times processor joined membership with
171 local processor. When processor fails and rejoins again, this
172 value is incremented.
173 status Status of processor. Can be one of joined and left.
175 resources.process.PID.*
178 Object created by applications using SAM with ConfDB integra‐
179 tion. It contains following keys:
180 recovery Recovery policy of process. Can be one of quit or
182 restart.
183 poll_period Value passed in sam_initialize as time_interval.
185 last_updated Last time when SAM received heartbeat from client.
187 state State of client. Can be one of failed, stopped, running
189 and waiting for quorum.
190 uidgid.*
193 Informations about users/groups which are allowed to do IPC con‐
194 nection to corosync. Objects can contain uid and gid keys, with
195 user and group allowed to connect corosync IPC.
196
199 corosync_overview(8), corosync.conf(5), corosync-objctl(8)
200corosync Man Page 04/05/2012 CONFDB_KEYS(8)