1
0
Files
2022-09-29 17:59:04 +03:00

469 lines
18 KiB
Plaintext

#
# $Revision: 2.21 $
#
# pmcd PMDA help file in the ASCII format
#
# lines beginning with a # are ignored
# lines beginning @ introduce a new entry of the form
# @ metric_name oneline-text
# help text goes
# here over multiple lines
# ...
#
# the metric_name is decoded against the default PMNS -- as a special case,
# a name of the form NNN.MM (for numeric NNN and MM) is interpreted as an
# instance domain identification, and the text describes the instance domain
#
# blank lines before the @ line are ignored
#
@ 2.1 Instance domain "pmloggers" from PMCD PMDA
This is the list of currently active pmlogger instances on the same machine as
this PMCD. The instance names are the process ids of the pmlogger instances.
The primary pmlogger has an extra instance with the instance name "primary" and
an instance id of zero (in addition to its normal process id instance).
@ 2.2 pmcd control register Instance Domain
One instance per pmcd control register.
The internal instance identifiers are the numbers 0 to 15.
The external instance names are he ASCII equivalent of the internal
instance identifiers.
@ 2.3 PMDA Instance Domain
One instance per PMDA managed by PMCD. The external and internal instance
identifiers are taken from the first two fields of the PMDA specification in
/etc/pmcd.conf.
@ 2.4 pmie Instance Domain
One instance per running pmie process. The internal and external instance
identifiers are the process ids of the pmie instances.
@ 2.5 buffer pool Instance Domain
The instances are as follows:
0012 12-byte buffers managed by __pmPoolAlloc and __pmPoolFree
0020 20-byte buffers managed by __pmPoolAlloc and __pmPoolFree
1024 1024-byte PDU buffers managed by __pmFindPDUBuf, __pmPinPDUBuf
and __pmUnpinPDUBuf
2048 2-Kbyte PDU buffers managed by __pmFindPDUBuf, __pmPinPDUBuf
and __pmUnpinPDUBuf
4096 3-Kbyte or 4-Kbyte PDU buffers managed by __pmFindPDUBuf,
__pmPinPDUBuf and __pmUnpinPDUBuf
8192 5-Kbyte, 6-Kbyte, 7-Kbyte or 8-Kbyte PDU buffers managed by
__pmFindPDUBuf, __pmPinPDUBuf and __pmUnpinPDUBuf
8192+ PDU buffers larger that 8-Kbyte managed by __pmFindPDUBuf,
__pmPinPDUBuf and __pmUnpinPDUBuf
@ pmcd.numagents Number of agents (PMDAs) currently connected to PMCD
The number of agents (PMDAs) currently connected to PMCD. This may differ
from the number of agents configured in /etc/pmcd/conf if agents have
terminated and/or been timed-out by PMCD.
@ pmcd.numclients Number of clients currently connected to PMCD
The number of connections open to client programs retrieving information from
PMCD.
@ pmcd.datasize Space allocated for PMCD and DSO agents' data segment (K)
This metric returns the amount of memory in kilobytes allocated for the data
segment of PMCD and any DSO agents (PMDAs) that it has loaded.
This is handy for tracing memory utilization (and leaks) in DSOs during
development.
@ pmcd.buf.alloc Allocated buffers in internal memory pools
This metric returns the number of allocated buffers for the various
buffer pools used by pmcd.
This is handy for tracing memory utilization (and leaks) in DSOs during
development.
@ pmcd.buf.free Free buffers in internal memory pools
This metric returns the number of free buffers for the various buffer
pools used by pmcd.
This is handy for tracing memory utilization (and leaks) in DSOs during
development.
@ pmcd.control.timeout Timeout interval for slow/hung agents (PMDAs)
PDU exchanges with agents (PMDAs) managed by PMCD are subject to timeouts
which detect and clean up slow or disfunctional agents. This metric returns
the current timeout period in seconds being used for the agents. If the value
is zero, timeouts are not being used. This corresponds to the -t option
described in the man page, pmcd(1).
It is possible to store a new timeout value into this metric. Storing zero
will turn off timeouts. Subsequent storing of a non-zero value will turn on
the timeouts again.
@ pmcd.control.debug Current value of PMCD debug flags
The current value of the PMCD debug flags. This is a bit-wise OR of the flags
described in the output of pmdbg -l. The PMCD-specific flags are:
DBG_TRACE_APPL0 2048 Trace agent & client I/O and termination
DBG_TRACE_APPL1 4096 Trace host access control
DBG_TRACE_APPL2 8192 Trace config file scanner and parser
It is possible to store values into this metric. Diagnostic output is written
to the PMCD log file (usually /var/adm/pcplog/pmcd.log).
Setting this metric to -1 terminates PMCD.
@ pmcd.pdu_in.total Total PDUs received by PMCD
Running total of all BINARY mode PDUs received by the PMCD from clients
and agents.
@ pmcd.pdu_in.error ERROR PDUs received by PMCD
Running total of BINARY mode ERROR PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.result RESULT PDUs received by PMCD
Running total of BINARY mode RESULT PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.profile PROFILE PDUs received by PMCD
Running total of BINARY mode PROFILE PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.fetch FETCH PDUs received by PMCD
Running total of BINARY mode FETCH PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.desc_req DESC_REQ PDUs received by PMCD
Running total of BINARY mode DESC_REQ PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.desc DESC PDUs received by PMCD
Running total of BINARY mode DESC PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.instance_req INSTANCE_REQ PDUs received by PMCD
Running total of BINARY mode INSTANCE_REQ PDUs received by the PMCD
from clients and agents.
@ pmcd.pdu_in.instance INSTANCE PDUs received by PMCD
Running total of BINARY mode INSTANCE PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.text_req TEXT_REQ PDUs received by PMCD
Running total of BINARY mode TEXT_REQ PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.text TEXT PDUs received by PMCD
Running total of BINARY mode TEXT PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.control_req CONTROL_REQ PDUs received by PMCD
Running total of BINARY mode CONTROL_REQ PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.datax DATAX PDUs received by PMCD
Running total of BINARY mode DATAX PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.creds CREDS PDUs received by PMCD
Running total of BINARY mode CREDS PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.pmns_ids PMNS_IDS PDUs received by PMCD
Running total of BINARY mode PMNS_IDS PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.pmns_names PMNS_NAMES PDUs received by PMCD
Running total of BINARY mode PMNS_NAMES PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.pmns_child PMNS_CHILD PDUs received by PMCD
Running total of BINARY mode PMNS_CHILD PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_in.pmns_traverse PMNS_TRAVERSE PDUs received by PMCD
Running total of BINARY mode PMNS_TRAVERSE PDUs received by the PMCD from
clients and agents.
@ pmcd.pdu_out.total Total PDUs sent by PMCD
Running total of all BINARY mode PDUs sent by the PMCD to clients and
agents.
@ pmcd.pdu_out.error ERROR PDUs sent by PMCD
Running total of BINARY mode ERROR PDUs sent by the PMCD to clients and
agents.
@ pmcd.pdu_out.result RESULT PDUs sent by PMCD
Running total of BINARY mode RESULT PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.profile PROFILE PDUs sent by PMCD
Running total of BINARY mode PROFILE PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.fetch FETCH PDUs sent by PMCD
Running total of BINARY mode FETCH PDUs sent by the PMCD to clients and
agents.
@ pmcd.pdu_out.desc_req DESC_REQ PDUs sent by PMCD
Running total of BINARY mode DESC_REQ PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.desc DESC PDUs sent by PMCD
Running total of BINARY mode DESC PDUs sent by the PMCD to clients and
agents.
@ pmcd.pdu_out.instance_req INSTANCE_REQ PDUs sent by PMCD
Running total of BINARY mode INSTANCE_REQ PDUs sent by the PMCD to
clients and agents.
@ pmcd.pdu_out.instance INSTANCE PDUs sent by PMCD
Running total of BINARY mode INSTANCE PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.text_req TEXT_REQ PDUs sent by PMCD
Running total of BINARY mode TEXT_REQ PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.text TEXT PDUs sent by PMCD
Running total of BINARY mode TEXT PDUs sent by the PMCD to clients and
agents.
@ pmcd.pdu_out.control_req CONTROL_REQ PDUs sent by PMCD
Running total of BINARY mode CONTROL_REQ PDUs sent by the PMCD to
clients and agents.
@ pmcd.pdu_out.datax DATAX PDUs sent by PMCD
Running total of BINARY mode DATAX PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.creds CREDS PDUs sent by PMCD
Running total of BINARY mode CREDS PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.pmns_ids PMNS_IDS PDUs sent by PMCD
Running total of BINARY mode PMNS_IDS PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.pmns_names PMNS_NAMES PDUs sent by PMCD
Running total of BINARY mode PMNS_NAMES PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.pmns_child PMNS_CHILD PDUs sent by PMCD
Running total of BINARY mode PMNS_CHILD PDUs sent by the PMCD to clients
and agents.
@ pmcd.pdu_out.pmns_traverse PMNS_TRAVERSE PDUs sent by PMCD
Running total of BINARY mode PMNS_TRAVERSE PDUs sent by the PMCD to clients
and agents.
@ pmcd.pmlogger.host host where active pmlogger is running
The fully qualified domain name of the host on which a pmlogger
instance is running.
The instance names are process ids of the active pmloggers. The
primary pmlogger has an extra instance with the instance name "primary"
and an instance id of zero (in addition to its normal process id
instance).
@ pmcd.pmlogger.port control port for active pmlogger
Each pmlogger instance has a port for receiving log control
information. This metric is a list of the active pmlogger control
ports on the same machine as this PMCD (i.e. the host identified in the
corresponding pmcd.pmlogger.host metric).
The instance names are process ids of the active pmloggers. The
primary pmlogger has an extra instance with the instance name "primary"
and an instance id of zero (in addition to its normal process id
instance).
@ pmcd.pmlogger.archive full pathname to archive basename for active pmlogger
The full pathname through the filesystem on the corresponding host
(pmcd.pmlogger.host) that is the base name for the archive log files.
The instance names are process ids of the active pmloggers. The
primary pmlogger has an extra instance with the instance name "primary"
and an instance id of zero (in addition to its normal process id
instance).
@ pmcd.pmlogger.pmcd_host host from which active pmlogger is fetching metrics
The fully qualified domain name of the host from which a pmlogger
instance is fetching metrics to be archived.
The instance names are process ids of the active pmloggers. The
primary pmlogger has an extra instance with the instance name "primary"
and an instance id of zero (in addition to its normal process id
instance).
@ pmcd.timezone local $TZ
Value for the $TZ environment variable where the PMCD is running.
Enables determination of "local" time for timestamps returned via
PMCD from a remote host.
@ pmcd.simabi MIPS subprogram interface model (SIM) ABI version of this PMCD
The version of the MIPS subprogram interface model (SIM) ABI used to create
this PMCD. The SIM is a subprogram call convention. PMCD is only able to
use DSO agents having the same SIM ABI version as the PMCD itself. The SIM
ABI version is used in constructing full pathnames from relative pathnames in
the PMCD configuration file (/etc/pmcd.conf).
@ pmcd.version PMCD version
@ pmcd.control.register a vector of registers that may be set by users
A vector of 16 32-bit registers that are identified by the instance
identifiers 0 through 15.
The register contents are initially zero, but may be subsequently
modified to be an arbitrary value using pmStore(3) or pmstore(1).
The values are not used internally, but rather act as a repository into
which operational information might be stored, and then exported to
modify the behavior of client programs, e.g. inhibit pmie(1) rule
firing, or trigger a status indicator. In this way,
pmcd.control.register acts like a primitive bulletin board.
Example use might be as follows
register[0] telephone no. of person assigned to current system problem
register[1] telephone no. of person assigned to current network problem
register[2] ORACLE database is down
register[3] backup in progress
register[4] shopping days to Christmas
@ pmcd.control.traceconn control PMCD connection event tracing
Set to 1 to enable PMCD event tracing for all connection-related
events for clients and PMDAs.
Set to 0 to disable PMCD connection event tracing.
@ pmcd.control.tracepdu control PMCD PDU event tracing
Set to 1 to enable PMCD event tracing for all PDUs sent and received
by PMCD.
Set to 0 to disable PMCD PDU event tracing.
@ pmcd.control.tracenobuf control buffering of PMCD event tracing
Set to 1 to enable unbuffered PMCD event tracing, where each event
is reported as it happens.
Set to 0 to enable buffering of PMCD event traces (this is the default),
and event traces will only be dumped or reported when an error occurs
or a value is stored into the PCP metric pmcd.control.dumptrace.
@ pmcd.control.tracebufs number of buffers for PMCD event tracing
Defaults to 20. May be changed dynamically.
@ pmcd.control.dumptrace force dump of PMCD event tracing buffers
Storing any value into this metric causes the PMCD event trace buffers
to be dumped to PMCD's log file.
@ pmcd.control.sighup force PMCD reset via SIGHUP
Storing any value into this metric causes PMCD to be reset by sending
itself a SIGHUP signal.
On reset (either by storing into pmcd.control.sighup or by sending
PMCD a SIGHUP directly), PMCD will restart any failed PMDAs and reload
the PMNS if it has been changed.
@ pmcd.control.dumpconn force dump of PMCD client connections
Storing any value into this metric causes the details of the current
PMCD client connections to be dumped to PMCD's log file.
@ pmcd.agent.type PMDA type
From /etc/pmcd.conf, this metric encodes the PMDA type as follows:
(x << 1) | y
where "x" is the IPC type between PMCD and the PMDA, i.e. 0 for DSO, 1
for socket or 2 for pipe, and "y" is the message passing style, i.e.
0 for binary or 1 for ASCII.
@ pmcd.agent.status PMDA status
This metric encodes the current status of each PMDA. The default value
is 0 if the PMDA is active.
Other values encode various degrees of PMDA difficulty in three bit
fields (bit 0 is the low-order bit) as follows:
bits 7..0
1 the PMDA is connected, but not yet "ready" to accept requests
from the PMDA
2 the PMDA has exited of its own accord
4 some error prevented the PMDA being started
8 PMCD stopped communication with the PMDA due to a protocol or
timeout error
bits 15..8
the exit() status from the PMDA
bits 23..16
the number of the signal that terminated the PMDA
@ pmcd.license PCP license capabilities at the PMCD host
A vector of bit fields that encode the PCP license capabilities
at the PMCD host.
Current bit-field values are:
1 PCP collector license
2 PCP monitor license
8 WebMeter license
@ pmcd.openfds highest PMCD file descriptor
The highest file descriptor index used by PMCD for a Client or PMDA connection.
@ pmcd.pmie.numrules number of rules being evaluated
The total number of rules being evaluated by each pmie process.
@ pmcd.pmie.eval.true count of pmie predicates evaluated to true
The predicate part of a pmie rule can be said to evaluate to either true,
false, or not known. This metric is a cumulative count of the number
of rules which have evaluated to true for each pmie instance.
@ pmcd.pmie.eval.false count of pmie predicates evaluated to false
The predicate part of a pmie rule can be said to evaluate to either true,
false, or not known. This metric is a cumulative count of the number
of rules which have evaluated to false for each pmie instance.
@ pmcd.pmie.eval.unknown count of pmie predicates not evaluated
The predicate part of a pmie rule can be said to evaluate to either true,
false, or not known. This metric is a cumulative count of the number
of rules which have not been successfully evaluated. This could be due to
not yet having sufficient values to evaluate the rule, or a metric fetch may
have been unsuccessful in retrieving current values for metrics required for
evaluation of the rule.
@ pmcd.pmie.eval.expected expected rate of rule evaluations
This is the expected rate of evaluation of pmie rules. The value is
calculated once when pmie starts, and is the number of pmie rules
divided by the average time interval over which they are to be evaluated.
@ pmcd.pmie.eval.actual count of actual rule evaluations
A cumulative count of the pmie rules which have been evaluated.
This value is incremented once for each evaluation of each rule.
@ pmcd.pmie.actions count of rules evaluating to true
A cumulative count of the evaluated pmie rules which have evaluated to true.
This value is incremented once each time an action is executed. This value
will always be less than or equal to pmcd.pmie.eval.true because predicates
which have evaluated to true may be suppressed in the action part of the
pmie rule, in which case this counter will not be incremented.
@ pmcd.pmie.configfile configuration file name
The full path in the filesystem to the configuration file containing the
rules being evaluated by each pmie instance.
If the configuration file was supplied on the standard input, then this
metric will have the value "<stdin>". If multiple configuration files
were given to pmie, then the value of this metric will be the first
configuration file specified.
@ pmcd.pmie.pmcd_host default hostname for pmie instance
The default host from which pmie is fetching metrics. This is either
the hostname given to pmie on the command line or the local host. Note
that this does not consider host names specified in the pmie configuration
file (these are considered non-default and can be more than one per pmie
instance). All daemon pmie instances started through pmie_check(1) will
have their default host passed in on their command line.
@ pmcd.pmie.logfile filename of pmie instance event log
The file to which each instance of pmie is writting events. No two pmie
instances can share the same log file. If no logfile was specified when
pmie was started, this metrics has the value "<none>". All daemon pmie
instances started through pmie_check(1) must have an associated log file.