public class StatisticsCollectorForOSX extends AbstractStatisticsCollector
Note: Unfortunately, the OSX command line reporting utilities do NOT provide access to IO Wait and do NOT break down IO into reads and writes. This means that we can not provide reporting for the following counters:
However, we are collecting the following which may be substituted to some extent: However, there is no substitute for IO Wait.https://sourceforge.net/apps/trac/bigdata/ticket/225AbstractStatisticsCollector.Options| Modifier and Type | Field and Description |
|---|---|
protected static int |
pid
The process identifier for this process (the JVM).
|
fullyQualifiedHostName, hostPathPrefix, interval, log, ps| Constructor and Description |
|---|
StatisticsCollectorForOSX(int interval,
String processName) |
| Modifier and Type | Method and Description |
|---|---|
CounterSet |
getCounters()
Return the counter hierarchy.
|
void |
start()
TODO Under OSX,
top provides more full featured reporting of
OS performance counters, including some IO counters and process specific
counters. |
void |
stop()
Stop collecting host performance data -- must be extended by the concrete
subclass.
|
addBasicServiceOrClientCounters, addGarbageCollectorMXBeanCounters, addMemoryPoolMXBeanCounters, addServiceProperties, getInterval, getMemoryCounterSet, getProcessName, installShutdownHook, kb2b, main, newInstancepublic StatisticsCollectorForOSX(int interval,
String processName)
interval - The interval at which the performance counters will be
collected in seconds.processName - The name of the process (or more typically its service
UUID) whose per-process performance counters are to
be collected.public void start()
top provides more full featured reporting of
OS performance counters, including some IO counters and process specific
counters. Unfortunately, it does NOT provide IO Wait and it does NOT
break out IO by reads and writes. It is also more work to parse its
output.
-l 0 turns on "logging" mode. The ZERO (0) indicates
infinite samples, rather than some finite number of samples.
-s [delay] sets the delay between reports.
-S adds reporting on swap space into the headers.
Swap: 545M + 479M free. Purgeable: 31M 21336(0) pages purged.Here is an example where the pid is NOT specified.
top -l 0 -s 60 -S Processes: 107 total, 3 running, 104 sleeping, 587 threads 2011/07/13 20:17:36 Load Avg: 2.24, 1.77, 1.53 CPU usage: 63.15% user, 36.84% sys, 0.0% idle SharedLibs: 5292K resident, 6092K data, 0B linkedit. MemRegions: 18245 total, 4813M resident, 30M private, 429M shared. PhysMem: 1027M wired, 5472M active, 671M inactive, 7171M used, 1019M free. VM: 263G vsize, 1042M framework vsize, 2911706(0) pageins, 2092534(0) pageouts. Swap: 545M + 479M free. Purgeable: 31M 21336(0) pages purged. Networks: packets: 20768691/24G in, 22255115/23G out. Disks: 3335765/133G read, 8511536/2051G written. PID COMMAND %CPU TIME #TH #WQ #PORTS #MREGS RPRVT RSHRD RSIZE VPRVT VSIZE PGRP PPID STATE UID FAULTS COW MSGSENT MSGRECV SYSBSD SYSMACH CSW PAGEINS USER 94188 mdworker 0.0 00:06.04 3 1 48+ 60+ 836K+ 16M+ 2488K+ 31M+ 2408M+ 94188 1 sleeping 89 32402+ 147+ 115066+ 37438+ 263505+ 79593+ 55853+ 42+ _spotlight 79486 Image Capture Ex 0.0 00:01.19 3 1 100+ 73+ 3052K+ 17M+ 7004K+ 40M+ 2672M+ 79486 865 sleeping 501 11662+ 174+ 126963+ 63971+ 26666+ 71795+ 16995+ 55+ bryan 79438 Preview 0.0 00:31.59 3 1 207+ 453+ 30M+ 67M+ 65M+ 77M+ 2851M+ 79438 865 sleeping 501 96609+ 1466+ 330566+ 162122+ 709661+ 553728+ 129001+ 801+ bryan 79114 TextEdit 0.0 00:04.80 3 2 121+ 193+ 6616K+ 28M+ 18M+ 44M+ 2751M+ 79114 865 sleeping 501 24588+ 870+ 211100+ 104015+ 52253+ 142782+ 43400+ 238+ bryan 66520- Skype 0.0 89:18.16 21 2 538+ 658+ 161M+ 78M+ 208M+ 270M+ 1266M+ 66520 865 sleeping 501 115403971+ 2498+ 12493857+ 6119421+ 10995806+ 28285072+ 17521056+ 3683+ bryanTop can be used to go after per-process statistics by specifying
-pid [pid]. It is also possible to request all processes
owned by a user with -user [user]. Here is an example where
the pid is specified.
top -l 0 -s 60 -S -pid 26866 Processes: 107 total, 3 running, 104 sleeping, 649 threads 2011/07/13 20:19:29 Load Avg: 1.14, 1.49, 1.45 CPU usage: 56.25% user, 43.75% sys, 0.0% idle SharedLibs: 5292K resident, 6060K data, 0B linkedit. MemRegions: 18439 total, 4806M resident, 30M private, 404M shared. PhysMem: 969M wired, 4768M active, 1366M inactive, 7102M used, 1089M free. VM: 263G vsize, 1042M framework vsize, 2911717(0) pageins, 2092534(0) pageouts. Swap: 545M + 479M free. Purgeable: 380K 22126(0) pages purged. Networks: packets: 20876610/24G in, 22362893/23G out. Disks: 3336157/133G read, 8512107/2051G written. PID COMMAND %CPU TIME #TH #WQ #PORTS #MREGS RPRVT RSHRD RSIZE VPRVT VSIZE PGRP PPID STATE UID FAULTS COW MSGSENT MSGRECV SYSBSD SYSMACH CSW PAGEINS USER 26866 java 0.0 20:29.90 137/1 1 5890+ 703+ 2823M+ 7944K+ 2863M+ 3281M+ 7450M+ 21051 21052 running 501 1975814+ 2352+ 9530469+ 138939+ 5625664+ 25326440+ 10828748+ 8030+ bryan
start in interface IStatisticsCollectorstart in class AbstractStatisticsCollectorpublic void stop()
AbstractStatisticsCollectorstop in interface IStatisticsCollectorstop in class AbstractStatisticsCollectorpublic CounterSet getCounters()
AbstractStatisticsCollectorAbstractProcessCollector.
Note: Subclasses MUST extend this method to initialize their own
counters.
TODO Why does this use the older synchronized pattern with a
shared AbstractStatisticsCollector.countersRoot object rather than returning a new object
per request? Check assumptions in the scale-out and local journal code
bases for this.
getCounters in interface ICounterSetAccessgetCounters in class AbstractStatisticsCollectorCopyright © 2006–2019 SYSTAP, LLC DBA Blazegraph. All rights reserved.