1PCP-ATOPSAR(1) General Commands Manual PCP-ATOPSAR(1)
2
6 pcp-atopsar - Advanced System Activity Report (pcp-atop related)
7
9 pcp [pcp options] atopsar [atop options] [-r file|date] [-h host] [-R
10 cnt] [-b hh:mm] [-e hh:mm]
11 pcp [pcp options] atopsar [atop options] interval [samples]
12
14 The pcp-atopsar program can be used to report statistics at the system
15 level.
16 In the first synopsis line (no sampling interval specified), pcp-atop‐
18 sar extracts data from a raw logfile that has been recorded previously
19 by pmlogger(1) (or via the -w option of the pcp-atop program).
20 You can specify the name of the logfile with the -r option of the pcp-
21 atopsar program. When a pmlogger daily logfile is used, named
22 $PCP_LOG_DIR/pmlogger/[host]/YYYYMMDD (where YYYYMMDD reflects the
23 date), the required date of the form YYYYMMDD can be specified with the
24 -r option instead of the filename, or the symbolic name 'y' can be used
25 for yesterday's daily logfile (this can be repeated so 'yyyy' indicates
26 the logfile of four days ago). If the -r option is not specified at
27 all, today's daily logfile is used by default.
28 By default, the hostname of the localhost will be used when resolving
29 pmlogger archives, however an alternative host can be specified using
30 the -h option.
31 The starting and ending times of the report can be defined using the
32 options -b and -e followed by a time argument of the form hh:mm.
33 In the second synopsis line, pcp-atopsar reads actual activity counters
35 from the kernel with the specified interval (in seconds) and the speci‐
36 fied number of samples (optionally). When pcp-atopsar is activated in
37 this way it immediately sends the output for every requested report to
38 standard output. If only one type of report is requested, the header
39 is printed once and after every interval seconds the statistical coun‐
40 ters are shown for that period. If several reports are requested, a
41 header is printed per sample followed by the statistical counters for
42 that period.
43 When invoked via the pcp(1) command, the PCPIntro(1) options -h/--host,
45 -a/--archive, -O/--origin, -s/--samples, -t/--interval, -Z/--timezone
46 and several other pcp options become indirectly available, see PCPIn‐
47 tro(1) for their descriptions.
48 Some generic flags can be specified to influence the behaviour of the
50 pcp-atopsar program:
51 -S By default the timestamp at the beginning of a line is suppressed
53 if more lines are shown for one interval. With this flag a time‐
54 stamp is given for every output-line (easier for post-processing).
55 -a By default certain resources as disks and network interfaces are
57 only shown when they were active during the interval. With this
58 flag all resources of a given type are shown, even if they were
59 inactive during the interval.
60 -x By default pcp-atopsar only uses colors if output is directed to a
62 terminal (window). These colors might indicate that a critical
63 occupation percentage has been reached (red) or has been almost
64 reached (cyan) for a particular resource. See the man-page of
65 atop for a detailed description of this feature (section COLORS).
66 With the flag -x the use of colors is suppressed unconditionally.
67 -C By default pcp-atopsar only uses colors if output is directed to a
69 terminal (window). These colors might indicate that a critical
70 occupation percentage has been reached (red) or has been almost
71 reached (cyan) for a particular resource. See the man-page of
72 atop for a detailed description of this feature (section COLORS).
73 With the flag -C colors will always be used, even if output is not
74 directed to a terminal.
75 -M Use markers at the end of a line to indicate that a critical occu‐
77 pation percentage has been reached ('*') or has been almost
78 reached ('+') for particular resources. The marker '*' is similar
79 to the color red and the marker '+' to the color cyan. See the
80 man-page of atop for a detailed description of these colors (sec‐
81 tion COLORS).
82 -H Repeat the header line within a report for every N detail lines.
84 The value of N is determined dynamically in case of output to a
85 tty/window (depending on the number of lines); for output to a
86 file or pipe this value is 23.
87 -R Summarize cnt samples into one sample. When the logfile contains
89 e.g. samples of 10 minutes, the use of the flag '-R 6' shows a
90 report with one sample for every hour.
91 Other flags are used to define which reports are required:
93 -A Show all possible reports.
95 -c Report about CPU utilization (in total and per cpu).
97 -g Report about GPU utilization (per GPU).
99 -p Report about processor-related matters, like load-averages and
101 hardware interrupts.
102 -P Report about processes.
104 -m Current memory- and swap-occupation.
106 -s Report about paging- and swapping-activity, and overcommitment.
108 -B Report about Pressure Stall Information (PSI).
110 -l Report about utilization of logical volumes.
112 -f Report about utilization of multiple devices.
114 -d Report about utilization of disks.
116 -n Report about NFS mounted filesystems on NFS client.
118 -j Report about NFS client activity.
120 -J Report about NFS server activity.
122 -i Report about the network interfaces.
124 -I Report about errors for network-interfaces.
126 -w Report about IP version 4 network traffic.
128 -W Report about errors for IP version 4 traffic.
130 -y General report about ICMP version 4 layer activity.
132 -Y Per-type report about ICMP version 4 layer activity.
134 -u Report about UDP version 4 network traffic.
136 -z Report about IP version 6 network traffic.
138 -Z Report about errors for IP version 6 traffic.
140 -k General report about ICMP version 6 layer activity.
142 -K Per-type report about ICMP version 6 layer activity.
144 -U Report about UDP version 6 network traffic.
146 -t Report about TCP network traffic.
148 -T Report about errors for TCP-traffic.
150 -h Report about Infiniband utilization.
152 -O Report about top-3 processes consuming most processor capacity.
154 This report is only available when using a log file (not when
155 specifying an interval).
156 -G Report about top-3 processes consuming most resident memory. This
158 report is only available when using a log file (not when specify‐
159 ing an interval).
160 -D Report about top-3 processes issueing most disk transfers. This
162 report is only available when using a log file (not when specify‐
163 ing an interval).
164 -N Report about top-3 processes issueing most IPv4/IPv6 socket trans‐
166 fers. This report is only available when using a log file (not
167 when specifying an interval).
168
170 The following additional PCP command line long options are also avail‐
171 able:
172 --align=align
174 Force the initial sample to be aligned on the boundary of a natu‐
175 ral time unit align. Refer to PCPIntro(1) for a complete descrip‐
176 tion of the syntax for align.
177 --archive=archive
179 Performance metric values are retrieved from the set of Perfor‐
180 mance Co-Pilot (PCP) archive log files identified by the argument
181 archive, which is a comma-separated list of names, each of which
182 may be the base name of an archive or the name of a directory con‐
183 taining one or more archives.
184 --finish=endtime
186 When reporting archived metrics, the report will be restricted to
187 those records logged before or at endtime. Refer to PCPIntro(1)
188 for a complete description of the syntax for endtime.
189 --host=host
191 Fetch performance metrics from pmcd(1) on host, rather than from
192 the default localhost.
193 --hostzone
195 Use the local timezone of the host that is the source of the per‐
196 formance metrics, as identified by either the --host or the --ar‐
197 chive options. The default is to use the timezone of the local
198 host.
199 --hotproc
201 Use the pmdaproc(1) hotproc metrics.
202 --interval=interval
204 Set the reporting interval to something other than the default 1
205 second. The interval argument follows the syntax described in
206 PCPIntro(1), and in the simplest form may be an unsigned integer
207 (the implied units in this case are seconds).
208 --samples=samples
210 The samples option defines the number of samples to be retrieved
211 and reported.
212 --start=starttime
214 When reporting archived metrics, the report will be restricted to
215 those records logged at or after starttime. Refer to PCPIntro(1)
216 for a complete description of the syntax for starttime.
217 --timezone=timezone
219 Use timezone for the date and time. Timezone is in the format of
220 the environment variable TZ as described in environ(7).
221 --version
223 Display version number and exit.
224
226 Depending on the requested report, a number of columns with output val‐
227 ues are produced. The values are mostly presented as a number of
228 events per second.
229 The output for the flag -c contains the following columns per cpu:
231 usr% Percentage of cpu-time consumed in user mode (program text)
233 for all active processes running with a nice value of zero
234 (default) or a negative nice value (which means a higher
235 priority than usual). The cpu consumption in user mode of
236 processes with a nice value larger than zero (lower prior‐
237 ity) is indicated in the nice%-column.
238 nice% Percentage of cpu time consumed in user mode (i.e. program
240 text) for all processes running witn a nice value larger
241 than zero (which means with a lower priority than average).
242 sys% Percentage of cpu time consumed in system mode (kernel
244 text) for all active processes. A high percentage usually
245 indicates a lot of system calls being issued.
246 irq% Percentage of cpu time consumed for handling of device
248 interrupts.
249 softirq% Percentage of cpu time consumed for soft interrupt han‐
251 dling.
252 steal% Percentage of cpu time stolen by other virtual machines
254 running on the same hardware.
255 guest% Percentage of cpu time used by other virtual machines run‐
257 ning on the same hardware (overlaps with usr%/nice%).
258 wait% Percentage of unused cpu time while at least one of the
260 processes in wait-state awaits completion of disk I/O.
261 idle% Percentage of unused cpu time because all processes are in
263 a wait-state but not waiting for disk-I/O.
264 The output for the flag -g contains the following columns per GPU:
266 busaddr GPU number and bus-ID (separated by '/').
268 gpubusy GPU busy percentage during interval.
270 membusy GPU memory busy percentage during interval, i.e. time to
272 issue read and write accesses on memory.
273 memocc Percentage of memory occupation at this moment.
275 memtot Total memory available.
277 memuse Used GPU memory at this moment.
279 gputype Type of GPU.
281 The output for the flag -p contains the following values:
283 pswch/s Number of process switches (also called context switches)
285 per second on this cpu. A process switch occurs at the
286 moment that an active thread (i.e. the thread using a cpu)
287 enters a wait state or has used its time slice completely;
288 another thread will then be chosen to use the cpu.
289 devintr/s Number of hardware interrupts handled per second on this
291 cpu.
292 clones/s The number of new threads started per second.
294 loadavg1 Load average reflecting the average number of threads in
296 the runqueue or in non-interruptible wait state (usually
297 waiting for disk or tape I/O) during the last minute.
298 loadavg5 Load average reflecting the average number of threads in
300 the runqueue or in non-interruptible wait state (usually
301 waiting for disk or tape I/O) during the last 5 minutes.
302 loadavg15 Load average reflecting the average number of threads in
304 the runqueue or in non-interruptible wait state (usually
305 waiting for disk or tape I/O) during the last 15 minutes.
306 The output for the flag -P contains information about the processes and
308 threads:
309 clones/s The number of new threads started per second.
311 pexit/s
313 curproc Total number of processes present in the system.
315 curzomb Number of zombie processes present in the system.
317 thrrun Total number of threads present in the system in state
319 'running'.
320 thrslpi Total number of threads present in the system in state
322 'interruptible sleeping'.
323 thrslpu Total number of threads present in the system in state
325 'uninterruptible sleeping'.
326 The output for the flag -m contains information about the memory- and
328 swap-utilization:
329 memtotal Total usable main memory size.
331 memfree Available main memory size at this moment (snapshot).
333 buffers Main memory used at this moment to cache metadata-blocks
335 (snapshot).
336 cached Main memory used at this moment to cache data-blocks (snap‐
338 shot).
339 dirty Amount of memory in the page cache that still has to be
341 flushed to disk at this moment (snapshot).
342 slabmem Main memory used at this moment for dynamically allocated
344 memory by the kernel (snapshot).
345 swptotal Total swap space size at this moment (snapshot).
347 swpfree Available swap space at this moment (snapshot).
349 The output for the flag -s contains information about the frequency of
351 swapping:
352 pagescan/s Number of scanned pages per second due to the fact that
354 free memory drops below a particular threshold.
355 swapin/s The number of memory-pages the system read from the swap-
357 device per second.
358 swapout/s The number of memory-pages the system wrote to the swap-
360 device per second.
361 commitspc The committed virtual memory space i.e. the reserved vir‐
363 tual space for all allocations of private memory space for
364 processes.
365 commitlim The maximum limit for the committed space, which is by
367 default swap size plus 50% of memory size. The kernel only
368 verifies whether the committed space exceeds the limit if
369 strict overcommit handling is configured (vm.overcom‐
370 mit_memory is 2).
371 The output for the flag -B contains the Pressure Stall Information
373 (PSI):
374 cs_10_60_300
376 Average pressure percentage over the last 10, 60 and 300
377 seconds for the category 'CPU some'.
378 ms_10_60_300
380 Average pressure percentage over the last 10, 60 and 300
381 seconds for the category 'memory some'.
382 mf_10_60_300
384 Average pressure percentage over the last 10, 60 and 300
385 seconds for the category 'memory full'.
386 is_10_60_300
388 Average pressure percentage over the last 10, 60 and 300
389 seconds for the category 'I/O some'.
390 if_10_60_300
392 Average pressure percentage over the last 10, 60 and 300
393 seconds for the category 'I/O full'.
394 The output for the flags -l (LVM), -f (MD), and -d (hard disk) contains
396 the following columns per active unit:
397 disk Name.
399 busy Busy-percentage of the unit (i.e. the portion of time that
401 the device was busy handling requests).
402 read/s Number of read-requests issued per second on this unit.
404 KB/read Average number of Kbytes transferred per read-request for
406 this unit.
407 writ/s Number of write-requests issued per second on this unit.
409 KB/writ Average number of Kbytes transferred per write-request for
411 this unit.
412 avque Average number of requests outstanding in the queue during
414 the time that the unit is busy.
415 avserv Average number of milliseconds needed by a request on this
417 unit (seek, latency and data-transfer).
418 The output for the flag -n contains information about activity on NFS
420 mounted filesystems (client):
421 mounted_device
423 Mounted device containing server name and server directory
424 being mounted.
425 physread/s Kilobytes data physically read from the NFS server by pro‐
427 cesses running on the NFS client.
428 KBwrite/s Kilobytes data physically written to the NFS server by pro‐
430 cesses running on the NFS client.
431 When the NFS filesystem was mounted during the interval,
432 the state 'M' is shown.
433 The output for the flag -j contains information about NFS client activ‐
435 ity:
436 rpc/s Number of RPC calls per second issued to NFS server(s).
438 rpcread/s Number of read RPC calls per second issued to NFS
440 server(s).
441 rpcwrite/s Number of write RPC calls per second issued to NFS
443 server(s).
444 retrans/s Number of retransmitted RPC calls per second.
446 autrefresh/s
448 Number of authorization refreshes per second.
449 The output for the flag -J contains information about NFS server activ‐
451 ity:
452 rpc/s Number of RPC calls per second received from NFS client(s).
454 rpcread/s Number of read RPC calls per second received from NFS
456 client(s).
457 rpcwrite/s Number of write RPC calls per second received from NFS
459 client(s).
460 MBcr/s Number of Megabytes per second returned to read requests by
462 clients.
463 MBcw/s Number of Megabytes per second passed in write requests by
465 clients.
466 nettcp/s Number of requests per second handled via TCP.
468 netudp/s Number of requests per second handled via UDP.
470 The output for the flag -i provides information about utilization of
472 network interfaces:
473 interf Name of interface.
475 busy Busy percentage for this interface. If the linespeed of
477 this interface could not be determined (e.g. for virtual
478 interfaces), a question mark is shown.
479 ipack/s Number of packets received from this interface per second.
481 opack/s Number of packets transmitted to this interface per second.
483 iKbyte/s Number of Kbytes received from this interface per second.
485 oKbyte/s Number of Kbytes transmitted via this interface per second.
487 imbps/s Effective number of megabits received per second.
489 ombps/s Effective number of megabits transmitted per second.
491 maxmbps/s Linespeed as number of megabits per second. If the line‐
493 speed could not be determined (e.g. virtual interfaces),
494 value 0 is shown.
495 The linespeed is followed by the indication 'f' (full
496 duplex) or 'h' (half duplex).
497 The output for the flag -I provides information about the failures that
499 were detected for network interfaces:
500 interf Name of interface.
502 ierr/s Number of bad packets received from this interface per sec‐
504 ond.
505 oerr/s Number of times that packet transmission to this interface
507 failed per second.
508 coll/s Number of collisions encountered per second while transmit‐
510 ting packets.
511 idrop/s Number of received packets dropped per second due to lack
513 of buffer-space in the local system.
514 odrop/s Number of transmitted packets dropped per second due to
516 lack of buffer-space in the local system.
517 iframe/s Number of frame alignment-errors encountered per second on
519 received packets.
520 ocarrier/s Number of carrier-errors encountered per second on trans‐
522 mitted packets.
523 The output for the flag -w provides information about the utilization
525 of the IPv4-layer (formal SNMP-names between brackets):
526 inrecv/s Number of IP datagrams received from interfaces per second,
528 including those received in error (ipInReceives).
529 outreq/s Number of IP datagrams that local higher-layer protocols
531 supplied to IP in requests for transmission per second
532 (ipOutRequests).
533 indeliver/s Number of received IP datagrams that have been successfully
535 delivered to higher protocol-layers per second (ipInDeliv‐
536 ers).
537 forward/s Number of received IP datagrams per second for which this
539 entity was not their final IP destination, as a result of
540 which an attempt was made to forward (ipForwDatagrams).
541 reasmok/s Number of IP datagrams successfully reassembled per second
543 (ipReasmOKs).
544 fragcreat/s Number of IP datagram fragments generated per second at
546 this entity (ipFragCreates).
547 The output for the flag -W provides information about the failures that
549 were detected in the IPv4-layer (formal SNMP-names between brackets):
550 in: dsc/s Number of input IP datagrams per second for which no prob‐
552 lems were encountered to prevent their continued processing
553 but that were discarded, e.g. for lack of buffer space
554 (ipInDiscards).
555 in: hder/s Number of input IP datagrams per second discarded due to
557 errors in the IP header (ipInHdrErrors).
558 in: ader/s Number of input IP datagrams per second discarded because
560 the IP address in the destination field was not valid to be
561 received by this entity (ipInAddrErrors).
562 in: unkp/s Number of inbound packets per second that were discarded
564 because of an unknown or unsupported protocol (ipInUnknown‐
565 Protos).
566 in: ratim/s Number of timeout-situations per second while other frag‐
568 ments were expected for successful reassembly (ipReasmTime‐
569 out).
570 in: rfail/s Number of failures detected per second by the IP reassembly
572 algorithm (ipReasmFails).
573 out: dsc/s Number of output IP datagrams per second for which no prob‐
575 lems were encountered to prevent their continued processing
576 but that were discarded, e.g. for lack of buffer space
577 (ipOutDiscards).
578 out: nrt/s Number of IP datagrams per second discarded because no
580 route could be found (ipOutNoRoutes).
581 The output for the flag -y provides information about the general uti‐
583 lization of the ICMPv4-layer and some information per type of ICMP-mes‐
584 sage (formal SNMP-names between brackets):
585 intot/s Number of ICMP messages (any type) received per second at
587 this entity (icmpInMsgs).
588 outtot/s Number of ICMP messages (any type) transmitted per second
590 from this entity (icmpOutMsgs).
591 inecho/s Number of ICMP Echo (request) messages received per second
593 (icmpInEchos).
594 inerep/s Number of ICMP Echo-Reply messages received per second
596 (icmpInEchoReps).
597 otecho/s Number of ICMP Echo (request) messages transmitted per sec‐
599 ond (icmpOutEchos).
600 oterep/s Number of ICMP Echo-Reply messages transmitted per second
602 (icmpOutEchoReps).
603 The output for the flag -Y provides information about other types of
605 ICMPv4-messages (formal SNMP-names between brackets):
606 ierr/s Number of ICMP messages received per second but determined
608 to have ICMP-specific errors (icmpInErrors).
609 isq/s Number of ICMP Source Quench messages received per second
611 (icmpInSrcQuenchs).
612 ird/s Number of ICMP Redirect messages received per second (icmp‐
614 InRedirects).
615 idu/s Number of ICMP Destination Unreachable messages received
617 per second (icmpInDestUnreachs).
618 ite/s Number of ICMP Time Exceeded messages received per second
620 (icmpOutTimeExcds).
621 oerr/s Number of ICMP messages transmitted per second but deter‐
623 mined to have ICMP-specific errors (icmpOutErrors).
624 osq/s Number of ICMP Source Quench messages transmitted per sec‐
626 ond (icmpOutSrcQuenchs).
627 ord/s Number of ICMP Redirect messages transmitted per second
629 (icmpOutRedirects).
630 odu/s Number of ICMP Destination Unreachable messages transmitted
632 per second (icmpOutDestUnreachs).
633 ote/s Number of ICMP Time Exceeded messages transmitted per sec‐
635 ond (icmpOutTimeExcds).
636 The output for the flag -u provides information about the utilization
638 of the UDPv4-layer (formal SNMP-names between brackets):
639 indgram/s Number of UDP datagrams per second delivered to UDP users
641 (udpInDatagrams).
642 outdgram/s Number of UDP datagrams transmitted per second from this
644 entity (udpOutDatagrams).
645 inerr/s Number of received UDP datagrams per second that could not
647 be delivered for reasons other than the lack of an applica‐
648 tion at the destination port (udpInErrors).
649 noport/s Number of received UDP datagrams per second for which there
651 was no application at the destination port (udpNoPorts).
652 The output for the flag -z provides information about the utilization
654 of the IPv6-layer (formal SNMP-names between brackets):
655 inrecv/s Number of input IPv6-datagrams received from interfaces per
657 second, including those received in error (ipv6IfStatsInRe‐
658 ceives).
659 outreq/s Number of IPv6-datagrams per second that local higher-layer
661 protocols supplied to IP in requests for transmission
662 (ipv6IfStatsOutRequests). This counter does not include
663 any forwarded datagrams.
664 inmc/s Number of multicast packets per second that have been
666 received by the interface (ipv6IfStatsInMcastPkts).
667 outmc/s Number of multicast packets per second that have been
669 transmitted to the interface (ipv6IfStatsOutMcastPkts).
670 indeliv/s Number of IP datagrams successfully delivered per second to
672 IPv6 user-protocols, including ICMP (ipv6IfStatsInDeliv‐
673 ers).
674 reasmok/s Number of IPv6 datagrams successfully reassembled per sec‐
676 ond (ipv6IfStatsReasmOKs).
677 fragcre/s Number of IPv6 datagram fragments generated per second at
679 this entity (ipv6IfStatsOutFragCreates).
680 The output for the flag -Z provides information about the failures that
682 were detected in the IPv6-layer (formal SNMP-names between brackets):
683 in: dsc/s Number of input IPv6 datagrams per second for which no
685 problems were encountered to prevent their continued pro‐
686 cessing but that were discarded, e.g. for lack of buffer
687 space (ipv6IfStatsInDiscards).
688 in: hder/s Number of input datagrams per second discarded due to
690 errors in the IPv6 header (ipv6IfStatsInHdrErrors).
691 in: ader/s Number of input datagrams per second discarded because the
693 IPv6 address in the destination field was not valid to be
694 received by this entity (ipv6IfStatsInAddrErrors).
695 in: unkp/s Number of locally-addressed datagrams per second that were
697 discarded because of an unknown or unsupported protocol
698 (ipv6IfStatsInUnknownProtos).
699 in: ratim/s Number of timeout-situations per second while other IPv6
701 fragments were expected for successful reassembly
702 (ipv6ReasmTimeout).
703 in: rfail/s Number of failures detected per second by the IPv6 reassem‐
705 bly-algorithm (ipv6IfStatsReasmFails).
706 out: dsc/s Number of output IPv6 datagrams per second for which no
708 problems were encountered to prevent their continued pro‐
709 cessing but that were discarded, e.g. for lack of buffer
710 space (ipv6IfStatsOutDiscards).
711 out: nrt/s Number of IPv6 datagrams per second discarded because no
713 route could be found (ipv6IfStatsInNoRoutes).
714 The output for the flag -k provides information about the general uti‐
716 lization of the ICMPv6-layer and some information per type of ICMP-mes‐
717 sage (formal SNMP-names between brackets):
718 intot/s Number of ICMPv6 messages (any type) received per second at
720 the interface (ipv6IfIcmpInMsgs).
721 outtot/s Number of ICMPv6 messages (any type) transmitted per second
723 from this entity (ipv6IfIcmpOutMsgs).
724 inerr/s Number of ICMPv6 messages received per second that had
726 ICMP-specific errors, such as bad ICMP checksums, bad
727 length, etc (ipv6IfIcmpInErrors).
728 innsol/s Number of ICMP Neighbor Solicit messages received per sec‐
730 ond (ipv6IfIcmpInNeighborSolicits).
731 innadv/s Number of ICMP Neighbor Advertisement messages received per
733 second (ipv6IfIcmpInNeighborAdvertisements).
734 otnsol/s Number of ICMP Neighbor Solicit messages transmitted per
736 second (ipv6IfIcmpOutNeighborSolicits).
737 otnadv/s Number of ICMP Neighbor Advertisement messages transmitted
739 per second (ipv6IfIcmpOutNeighborAdvertisements).
740 The output for the flag -K provides information about other types of
742 ICMPv6-messages (formal SNMP-names between brackets):
743 iecho/s Number of ICMP Echo (request) messages received per second
745 (ipv6IfIcmpInEchos).
746 ierep/s Number of ICMP Echo-Reply messages received per second
748 (ipv6IfIcmpInEchoReplies).
749 oerep/s Number of ICMP Echo-Reply messages transmitted per second
751 (ipv6IfIcmpOutEchoReplies).
752 idu/s Number of ICMP Destination Unreachable messages received
754 per second (ipv6IfIcmpInDestUnreachs).
755 odu/s Number of ICMP Destination Unreachable messages transmitted
757 per second (ipv6IfIcmpOutDestUnreachs).
758 ird/s Number of ICMP Redirect messages received per second
760 (ipv6IfIcmpInRedirects).
761 ord/s Number of ICMP Redirect messages transmitted per second
763 (ipv6IfIcmpOutRedirect).
764 ite/s Number of ICMP Time Exceeded messages received per second
766 (ipv6IfIcmpInTimeExcds).
767 ote/s Number of ICMP Time Exceeded messages transmitted per sec‐
769 ond (ipv6IfIcmpOutTimeExcds).
770 The output for the flag -U provides information about the utilization
772 of the UDPv6-layer (formal SNMP-names between brackets):
773 indgram/s Number of UDPv6 datagrams per second delivered to UDP users
775 (udpInDatagrams),
776 outdgram/s Number of UDPv6 datagrams transmitted per second from this
778 entity (udpOutDatagrams),
779 inerr/s Number of received UDPv6 datagrams per second that could
781 not be delivered for reasons other than the lack of an
782 application at the destination port (udpInErrors).
783 noport/s Number of received UDPv6 datagrams per second for which
785 there was no application at the destination port (udpNo‐
786 Ports).
787 The output for the flag -t provides information about the utilization
789 of the TCP-layer (formal SNMP-names between brackets):
790 insegs/s Number of received segments per second, including those
792 received in error (tcpInSegs).
793 outsegs/s Number of transmitted segments per second, excluding those
795 containing only retransmitted octets (tcpOutSegs).
796 actopen/s Number of active opens per second that have been supported
798 by this entity (tcpActiveOpens).
799 pasopen/s Number of passive opens per second that have been supported
801 by this entity (tcpPassiveOpens).
802 nowopen Number of connections currently open (snapshot), for which
804 the state is either ESTABLISHED or CLOSE-WAIT (tcpCur‐
805 rEstab).
806 The output for the flag -T provides information about the failures that
808 were detected in the TCP-layer (formal SNMP-names between brackets):
809 inerr/s Number of received segments per second received in error
811 (tcpInErrs).
812 retrans/s Number of retransmitted segments per second (tcpRe‐
814 transSegs).
815 attfail/s Number of failed connection attempts per second that have
817 occurred at this entity (tcpAttemptFails).
818 estabreset/s
820 Number of resets per second that have occurred at this
821 entity (tcpEstabResets).
822 outreset/s Number of transmitted segments per second containing the
824 RST flag (tcpOutRsts).
825 The output for the flag -h provides information about utilization of
827 Infiniband ports:
828 controller Name of controller.
830 port Controller port.
832 busy Busy percentage for this port.
834 ipack/s Number of packets received from this port per second.
836 opack/s Number of packets transmitted to this port per second.
838 igbps/s Effective number of gigabits received per second.
840 ogbps/s Effective number of gigabits transmitted per second.
842 maxgbps/s Maximum rate as number of gigabits per second.
844 lanes Number of lanes.
846 The output for the flag -O provides information about the top-3 of pro‐
848 cesses with the highest processor consumption:
849 pid Process-id (if zero, the process has exited while the pid
851 could not be determined).
852 command The name of the process.
854 cpu% The percentage of cpu-capacity being consumed. This value
856 can exceed 100% for a multithreaded process running on a
857 multiprocessor machine.
858 The output for the flag -G provides information about the top-3 of pro‐
860 cesses with the highest memory consumption:
861 pid Process-id (if zero, the process has exited while the pid
863 could not be determined).
864 command The name of the process.
866 mem% The percentage of resident memory-utilization by this
868 process.
869 The output for the flag -D provides information about the top-3 of pro‐
871 cesses that issue the most read and write accesses to disk:
872 pid Process-id (if zero, the process has exited while the pid
874 could not be determined).
875 command The name of the process.
877 dsk% The percentage of read and write accesses related to the
879 total number of read and write accesses issued on disk by
880 all processes, so a high percentage does not imply a high
881 disk load on system level.
882 The output for the flag -N provides information about the top-3 of pro‐
884 cesses that issue the most socket transfers for IPv4/IPv6:
885 pid Process-id (if zero, the process has exited while the pid
887 could not be determined).
888 command The name of the process.
890 net% The percentage of socket transfers related to the total
892 number of transfers issued by all processes, so a high per‐
893 centage does not imply a high network load on system level.
894
896 To see today's cpu-activity so far (supposed that atop is logging in
897 the background):
898 pcp-atopsar
900 To see the memory occupation for June 5, 2018 between 10:00 and 12:30
902 (supposed that pmlogger has been logging daily in the background on
903 host acme.com):
904 pcp-atopsar -m -r $PCP_LOG_DIR/pmlogger/acme.com/20180605 -b 10:00 -e
906 12:30
907 or
909 pcp-atopsar -m -r 20180605 -b 10:00 -e 12:30
911 or, suppose it is June 8, 2018 at this moment
913 pcp-atopsar -m -r yyy -b 10:00 -e 12:30
915 Write a logfile with atop to record the system behaviour for 30 minutes
917 (30 samples of one minute) and produce all available reports after‐
918 wards:
919 pcp-atop -w /tmp/atoplog 60 30
921 pcp-atopsar -A -r /tmp/atoplog
923 To watch TCP activity evolve for ten minutes (10 samples with sixty
925 seconds interval):
926 pcp-atopsar -t 60 10
928 To watch the header-lines ('_' as last character) of all reports with
930 only the detail-lines showing critical resource consumption (marker '*'
931 or '+' as last character):
932 pcp-atopsar -AM | grep '[_*+]$'
934
936 /etc/atoprc
937 Configuration file containing system-wide default values (mainly
938 flags). See related man-page.
939 ~/.atoprc
941 Configuration file containing personal default values (mainly
942 flags). See related man-page.
943 $PCP_LOG_DIR/pmlogger/HOST/YYYYMMDD
945 Daily data file, where YYYYMMDD are digits representing the date,
946 and HOST is the hostname of the machine being logged.
947
949 Environment variables with the prefix PCP_ are used to parameterize the
950 file and directory names used by PCP. On each installation, the file
951 /etc/pcp.conf contains the local values for these variables. The
952 $PCP_CONF variable may be used to specify an alternative configuration
953 file, as described in pcp.conf(5).
954 For environment variables affecting PCP tools, see pmGetOptions(3).
956
958 PCPIntro(1), pcp(1), pcp-atop(1), mkaf(1), pmlogger(1), pmlog‐
959 ger_daily(1) and pcp-atoprc(5).
960Performance Co-Pilot PCP PCP-ATOPSAR(1)