1PMLOGGER_DAILY(1) General Commands Manual PMLOGGER_DAILY(1)
2
6 pmlogger_daily - administration of Performance Co-Pilot archive files
7
9 $PCP_BINADM_DIR/pmlogger_daily [-DEfKMNoprRVzZ?] [-c control] [-k
10 time] [-l logfile] [-m addresses] [-s size] [-t want] [-x time] [-X
11 program] [-Y regex]
12
14 pmlogger_daily and the related pmlogger_check(1) tools along with asso‐
15 ciated control files (see pmlogger.control(5)) may be used to create a
16 customized regime of administration and management for historical ar‐
17 chives of performance data within the Performance Co-Pilot (see PCPIn‐
18 tro(1)) infrastructure.
19 pmlogger_daily is intended to be run once per day, preferably in the
21 early morning, as soon after midnight as practicable. Its task is to
22 aggregate, rotate and perform general housekeeping one or more sets of
23 PCP archives.
24 To accommodate the evolution of PMDAs and changes in production logging
26 environments, pmlogger_daily is integrated with pmlogrewrite(1) to al‐
27 low optional and automatic rewriting of archives before merging. If
28 there are global rewriting rules to be applied across all archives men‐
29 tioned in the control file(s), then create the directory
30 $PCP_SYSCONF_DIR/pmlogrewrite and place any pmlogrewrite(1) rewriting
31 rules in this directory. For rewriting rules that are specific to only
32 one family of archives, use the directory name from the control file(s)
33 - i.e. the fourth field - and create a file, or a directory, or a sym‐
34 bolic link named pmlogrewrite within this directory and place the re‐
35 quired rewriting rule(s) in the pmlogrewrite file or in files within
36 the pmlogrewrite subdirectory. pmlogger_daily will choose rewriting
37 rules from the archive directory if they exist, else rewriting rules
38 from $PCP_SYSCONF_DIR/pmlogrewrite if that directory exists, else no
39 rewriting is attempted.
40 As an alternate mechanism, if the file $PCP_LOG_DIR/pmlogger/.Nee‐
42 dRewrite exists when pmlogger_daily starts then this is treated the
43 same as specifying -R on the command line and $PCP_LOG_DIR/pmlog‐
44 ger/.NeedRewrite will be removed once all the rewriting has been done.
45
47 -c control, --control=control
48 Both pmlogger_daily and pmlogger_check(1) are controlled by PCP
49 logger control file(s) that specifies the pmlogger instances to be
50 managed. The default control file is $PCP_PMLOGGERCONTROL_PATH,
51 but an alternate may be specified using the -c option. If the di‐
52 rectory $PCP_PMLOGGERCONTROL_PATH.d (or control.d from the -c op‐
53 tion) exists, then the contents of any additional control files
54 therein will be appended to the main control file (which must ex‐
55 ist).
56 -D, --noreport
58 Do not perform the conditional pmlogger_daily_report(1) processing
59 as described below.
60 -E, --expunge
62 This option causes pmlogger_daily to pass the -E flag to pmlog‐
63 ger_merge(1) in order to expunge metrics with metadata inconsis‐
64 tencies and continue rather than fail. This is intended for auto‐
65 mated daily archive rotation where it is highly desirable for
66 unattended daily archive merging, rewriting and compression to
67 succeed. For further details, see pmlogger_merge(1) and descrip‐
68 tion for the -x flag in pmlogextract(1).
69 -f, --force
71 This option forces pmlogger_daily to attempt compression actions.
72 Using this option in production is not recommended.
73 -k time, --discard=time
75 After some period, old PCP archives are discarded. time is a time
76 specification in the syntax of find-filter(1), so DD[:HH[:MM]].
77 The optional HH (hours) and MM (minutes) parts are 0 if not speci‐
78 fied. By default the time is 14:0:0 or 14 days, but may be
79 changed using this option.
80 Some special values are recognized for the time, namely 0 to keep
82 no archives beyond the the ones being currently written by pmlog‐
83 ger(1), and forever or never to prevent any archives being dis‐
84 carded.
85 The time can also be set using the $PCP_CULLAFTER variable, set in
87 either the environment or in a control file. If both $PCP_CUL‐
88 LAFTER and -k specify different values for time then the environ‐
89 ment variable value is used and a warning is issued, i.e. if
90 $PCP_CULLAFTER is set in the control file, it overrides -k given
91 on the command line.
92 Note that the semantics of time are that it is measured from the
94 time of last modification of each archive, and not from the origi‐
95 nal archive creation date. This has subtle implications for com‐
96 pression (see below) - the compression process results in the cre‐
97 ation of new archive files which have new modification times. In
98 this case, the time period (re)starts from the time of compres‐
99 sion.
100 -K When this option is specified for pmlogger_daily then only the
102 compression tasks are attempted, so no pmlogger rotation, no
103 culling, no rewriting, etc. When -K is used and a period of 0 is
104 in effect (from -x on the command line or $PCP_COMPRESSAFTER in
105 the environment or via the control file) this is intended for en‐
106 vironments where compression of archives is desired before the
107 scheduled daily processing happens. To achieve this, once pmlog‐
108 ger_check(1) has completed regular processing, it calls pmlog‐
109 ger_daily with just the -K option. Provided $PCP_COMPRESSAFTER is
110 set to 0 along with any other required compression options to
111 match the scheduled invocation of pmlogger_daily, then this will
112 compress all volumes except the ones being currently written by
113 pmlogger(1). If $PCP_COMPRESSAFTER is set to a value greater than
114 zero, then manually running pmlogger_daily with the -x option may
115 be used to compress volumes that are younger than the $PCP_COM‐
116 PRESSAFTER time. This may be used to reclaim filesystem space by
117 compressing volumes earlier than they would have otherwise been
118 compressed. Note that since the default value of $PCP_COMPRES‐
119 SAFTER is 0 days, the -x option has no effect unless the control
120 file has been edited and $PCP_COMPRESSAFTER has been set to a
121 value greater than 0.
122 -l file, --logfile=file
124 In order to ensure that mail is not unintentionally sent when
125 these scripts are run from cron(8) or systemd(1) diagnostics are
126 always sent to log files. By default, this file is
127 $PCP_LOG_DIR/pmlogger/pmlogger_daily.log but this can be changed
128 using the -l option. If this log file already exists when the
129 script starts, it will be renamed with a .prev suffix (overwriting
130 any log file saved earlier) before diagnostics are generated to
131 the log file. The -l and -t options cannot be used together.
132 -m addresses, --mail=addresses
134 Use of this option causes pmlogger_daily to construct a summary of
135 the ``notices'' file entries which were generated in the last 24
136 hours, and e-mail that summary to the set of space-separated ad‐
137 dresses. This daily summary is stored in the file
138 $PCP_LOG_DIR/NOTICES.daily, which will be empty when no new ``no‐
139 tices'' entries were made in the previous 24 hour period.
140 -M This option may be used to disable archive merging (or renaming)
142 and rewriting (-M implies -r). This is most useful in cases where
143 the archives are being incrementally copied to a remote reposi‐
144 tory, e.g. using rsync(1). Merging, renaming and rewriting all
145 risk an increase in the synchronization load, especially immedi‐
146 ately after pmlogger_daily has run, so -M may be useful in these
147 cases.
148 -N, --showme
150 This option enables a ``show me'' mode, where the programs actions
151 are echoed, but not executed, in the style of ``make -n''. Using
152 -N in conjunction with -V maximizes the diagnostic capabilities
153 for debugging.
154 -o By default all possible archives will be merged. This option re‐
156 instates the old behaviour in which only yesterday's archives will
157 be considered as merge candidates. In the special case where only
158 a single input archive needs to be merged, pmlogmv(1) is used to
159 rename the archive, otherwise pmlogger_merge(1) is used to merge
160 all of the archives for a single host and a single day into a new
161 PCP archive and the individual archives are removed.
162 -p If this option is specified for pmlogger_daily then the status of
164 the daily processing is polled and if the daily pmlogger(1) rota‐
165 tion, culling, rewriting, compressing, etc. has not been done in
166 the last 24 hours then it is done now. The intent is to have pm‐
167 logger_daily called regularly with the -p option (at 30 mins past
168 the hour, every hour in the default cron(8) set up) to ensure
169 daily processing happens as soon as possible if it was missed at
170 the regularly scheduled time (which is 00:10 by default), e.g. if
171 the system was down or suspended at that time. With this option
172 pmlogger_daily simply exits if the previous day's processing has
173 already been done. Note that this option is not used on platforms
174 supporting systemd(1) because the pmlogger_daily.timer service
175 unit specifies a timer setting with Persistent=true. The -K and
176 -p options to pmlogger_daily are mutually exclusive.
177 -r, --norewrite
179 This command line option acts as an override and prevents all ar‐
180 chive rewriting with pmlogrewrite(1) independent of the presence
181 of any rewriting rule files or directories.
182 -R, --rewriteall
184 Sometimes PMDA changes require all archives to be rewritten, not
185 just the ones involved in any current merging. This is required
186 for example after a PCP upgrade where a new version of an existing
187 PMDA has revised metadata. The -R command line forces this uni‐
188 versal-style of rewriting. The -R option to pmlogger_daily is mu‐
189 tually exclusive with both the -r and -M options.
190 -s size, --rotate=size
192 If the PCP ``notices'' file ($PCP_LOG_DIR/NOTICES) is larger than
193 20480 bytes, pmlogger_daily will rename the file with a ``.old''
194 suffix, and start a new ``notices'' file. The rotate threshold
195 may be changed from 20480 to size bytes using the -s option.
196 -t period
198 To assist with debugging or diagnosing intermittent failures the
199 -t option may be used. This will turn on very verbose tracing
200 (-VV) and capture the trace output in a file named
201 $PCP_LOG_DIR/pmlogger/daily.datestamp.trace, where datestamp is
202 the time pmlogger_daily was run in the format YYYYMMDD.HH.MM. In
203 addition, the period argument will ensure that trace files created
204 with -t will be kept for period days and then discarded.
205 -V, --verbose
207 The output from the cron execution of the scripts may be extended
208 using the -V option to the scripts which will enable verbose trac‐
209 ing of their activity. By default the scripts generate no output
210 unless some error or warning condition is encountered. A second
211 -V increases the verbosity. Using -N in conjunction with -V maxi‐
212 mizes the diagnostic capabilities for debugging.
213 -x time, --compress-after=time
215 Archive data files can optionally be compressed after some period
216 to conserve disk space. This is particularly useful for large
217 numbers of pmlogger processes under the control of pmlogger_daily.
218 time is a time specification in the syntax of find-filter(1), so
220 DD[:HH[:MM]]. The optional HH (hours) and MM (minutes) parts are
221 0 if not specified.
222 Some special values are recognized for the time, namely 0 to apply
224 compression as soon as possible, and forever or never to prevent
225 any compression being done.
226 If transparent_decompress is enabled when libpcp was built (can be
228 checked with the pmconfig(1) -L option), then the default behav‐
229 iour is compression ``as soon as possible''. Otherwise the de‐
230 fault behaviour is to not compress files (which matches the his‐
231 torical default behaviour in earlier PCP releases).
232 The time can also be set using the $PCP_COMPRESSAFTER variable,
234 set in either the environment or in a control file. If both
235 $PCP_COMPRESSAFTER and -x specify different values for time then
236 the environment variable value is used and a warning is issued.
237 For important other detailed notes concerning volume compression,
238 see the -K and -k options (above).
239 -X program, --compressor=program
241 This option specifies the program to use for compression - by de‐
242 fault this is xz(1). The environment variable $PCP_COMPRESS may
243 be used as an alternative mechanism to define program. If both
244 $PCP_COMPRESS and -X specify different compression programs then
245 the environment variable value is used and a warning is issued.
246 -Y regex, --regex=regex
248 This option allows a regular expression to be specified causing
249 files in the set of files matched for compression to be omitted -
250 this allows only the data file to be compressed, and also prevents
251 the program from attempting to compress it more than once. The
252 default regex is ".(index|Z|gz|bz2|zip|xz|lzma|lzo|lz4)$" - such
253 files are filtered using the -v option to egrep(1). The environ‐
254 ment variable $PCP_COMPRESSREGEX may be used as an alternative
255 mechanism to define regex. If both $PCP_COMPRESSREGEX and -Y
256 specify different values for regex then the environment variable
257 value is used and a warning is issued.
258 -z This option causes pmlogger_daily to not ``re-exec'', see pmlog‐
260 ger(1), when it would otherwise choose to do so and is intended
261 only for QA testing.
262 -Z This option causes pmlogger_daily to ``re-exec'', see pmlogger(1),
264 whenever that is possible and is intended only for QA testing.
265 -?, --help
267 Display usage message and exit.
268
270 Additionally pmlogger_daily supports the following ``hooks'' to allow
271 auxiliary operations to be performed at key points in the daily pro‐
272 cessing of the archives. These callbacks are controlled via variables
273 that may be set in the environment or via the control file.
274 Note that merge callbacks and autosaving described below are not en‐
276 abled when only compression tasks are being attempted, i.e. when -K
277 command line option is used.
278 All of the callback script execution and the autosave file moving will
280 be executed as the non-privileged user ``pcp'' and group ``pcp'', so
281 appropriate permissions may need to have been set up in advance.
282 $PCP_MERGE_CALLBACK
284 As each day's archive is created by merging and before any com‐
285 pression takes place, if $PCP_MERGE_CALLBACK is defined, then it
286 is assumed to be a script that will be called with one argument
287 being the name of the archive (stripped of any suffixes), so some‐
288 thing of the form /some/directory/path/YYYYMMDD. The script needs
289 to be either a full pathname, or something that will be found on
290 the shell's $PATH . The callback script will be run in the fore‐
291 ground, so pmlogger_daily will wait for it to complete.
292 If the control file contains more than one $PCP_MERGE_CALLBACK
294 specification then these will be run serially in the order they
295 appear in the control file. If $PCP_MERGE_CALLBACK is defined in
296 the environment when pmlogger_daily is run, this is treated as
297 though this option was the first in the control file, i.e. it will
298 be run before any merge callbacks mentioned in the control file.
299 If the pcp-zeroconf packages is installed, then a special merge
301 callback is added to call pmlogger_daily_report(1) first, before
302 any other merge callback options. Refer to pmlogger_daily_re‐
303 port(1) for an explanation of the pcp-zeroconf requirements.
304 If pmlogger_daily is in ``catch up'' mode (more than one day's
306 worth of archives need to be combined) then each call back is exe‐
307 cuted once for each day's archive that is generated.
308 A typical use might be to produce daily reports from the PCP ar‐
310 chive which needs to wait until the archive has been created, but
311 is more efficient if it is done before any potential compression
312 of the archive.
313 $PCP_COMPRESS_CALLBACK
315 If pmlogger_daily is run with -x 0 or $PCP_COMPRESSAFTER=0, then
316 compression is done immediately after merging. As each day's ar‐
317 chive is compressed, if $PCP_COMPRESS_CALLBACK is defined, then it
318 is assumed to be a script that will be called with one argument
319 being the name of the archive (stripped of any suffixes), so some‐
320 thing of the form /some/directory/path/YYYYMMDD. The script needs
321 to be either a full pathname, or something that will be found on
322 the shell's $PATH . The callback script will be run in the fore‐
323 ground, so pmlogger_daily will wait for it to complete.
324 If the control file contains more than one $PCP_COMPRESS_CALLBACK
326 specification then these will be run serially in the order they
327 appear in the control file. If $PCP_COMPRESS_CALLBACK is defined
328 in the environment when pmlogger_daily is run, this is treated as
329 though this option was the first in the control file, i.e. it will
330 be run first.
331 If pmlogger_daily is in ``catch up'' mode (more than one day's
333 worth of archives need to be compressed) then each call back is
334 executed once for each day's archive that is compressed.
335 A typical use might be to keep recent archives in uncompressed
337 form for efficient querying, but move the older archives to some
338 other storage location once the compression has been done.
339 $PCP_AUTOSAVE_DIR
341 Once the merging and possible compression has been done by pmlog‐
342 ger_daily, if $PCP_AUTOSAVE_DIR is defined then all of the physi‐
343 cal files that make up one day's archive will be moved (autosaved)
344 to the directory specified by $PCP_AUTOSAVE_DIR.
345 The basename of the archive is used to set the reserved words
347 DATEYYYY (year), DATEMM (month) and DATEDD (day) and these (along
348 with LOCALHOSTNAME) may appear literally in $PCP_AUTOSAVE_DIR, and
349 will be substituted at execution time to generate the destination
350 directory name. For example:
351 $PCP_AUTOSAVE_DIR=/gpfs/LOCALHOSTNAME/DATEYYYY/DATEMM-DATEDD
352 Note that these ``date'' reserved words correspond to the date on
354 which the archive data was collected, not the date that pmlog‐
355 ger_daily was run.
356 If $PCP_AUTOSAVE_DIR (after LOCALHOSTNAME and ``date'' substitu‐
358 tion) does not exist then pmlogger_daily will attempt to create it
359 (along with any parent directories that do not exist). Just be
360 aware that this directory creation runs under the uid of the user
361 ``pcp'', so directories along the path to $PCP_AUTOSAVE_DIR may
362 need to be writeable by this non-root user.
363 By ``move'' the archives we mean a paranoid checksum-copy-check‐
365 sum-remove (using the -c option for pmlogmv(1)) that will bail if
366 the copy fails or the checksums do not match (the archives are im‐
367 portant so we cannot risk something like a full filesystem or a
368 permissions issue messing with the copy process).
369 If pmlogger_daily is in ``catch up'' mode (more than one day's
371 worth of archives need to be combined) then the archives for more
372 than one day could be copied in this step.
373 A typical use might be to create PCP archives on a local filesys‐
375 tem initially, then once all the data for a single day has been
376 collected and merged, migrate that day's archive to a shared
377 filesystem or a remote filesystem. This may allow automatic
378 backup to off-site storage and/or reduce the number of I/O opera‐
379 tions and filesystem metadata operations on the (potentially
380 slower) non-local filesystem.
381
383 Refer to pmlogger.control(5) for a description of the contol file(s)
384 that are used to control which pmlogger instances and which archives
385 are managed by pmlogger_check and pmlogger_daily(1).
386
388 $PCP_VAR_DIR/config/pmlogger/config.default
389 default pmlogger configuration file location for the local primary
390 logger, typically generated automatically by pmlogconf(1).
391 $PCP_ARCHIVE_DIR/<hostname>
393 default location for archives of performance information collected
394 from the host hostname
395 $PCP_ARCHIVE_DIR/<hostname>/lock
397 transient lock file to guarantee mutual exclusion during pmlogger
398 administration for the host hostname - if present, can be safely
399 removed if neither pmlogger_daily nor pmlogger_check(1) are run‐
400 ning
401 $PCP_ARCHIVE_DIR/<hostname>/Latest
403 PCP archive folio created by mkaf(1) for the most recently
404 launched archive containing performance metrics from the host
405 hostname
406 $PCP_LOG_DIR/NOTICES
408 PCP ``notices'' file used by pmie(1) and friends
409 $PCP_LOG_DIR/pmlogger/pmlogger_daily.log
411 if the previous execution of pmlogger_daily produced any output it
412 is saved here. The normal case is no output in which case the
413 file does not exist.
414 $PCP_ARCHIVE_DIR/<hostname>/SaveLogs
416 if this directory exists, then the log file from the -l argument
417 of a newly launched pmlogger(1) for hostname will be linked into
418 this directory with the name archive.log where archive is the
419 basename of the associated pmlogger(1) PCP archive files. This
420 allows the log file to be inspected at a later time, even if sev‐
421 eral pmlogger(1) instances for hostname have been launched in the
422 interim. Because the PCP archive management tools run under the
423 uid of the user ``pcp'', $PCP_ARCHIVE_DIR/<hostname>/SaveLogs typ‐
424 ically needs to be owned by the user ``pcp''.
425 $PCP_LOG_DIR/pmlogger/.NeedRewrite
427 if this file exists, then this is treated as equivalent to using
428 -R on the command line and the file will be removed once all
429 rewriting has been done.
430
432 Environment variables with the prefix PCP_ are used to parameterize the
433 file and directory names used by PCP. On each installation, the file
434 /etc/pcp.conf contains the local values for these variables. The
435 $PCP_CONF variable may be used to specify an alternative configuration
436 file, as described in pcp.conf(5).
437
439 Earlier versions of pmlogger_daily used find(1) to locate files for
440 compressing or culling and the -k and -x options took only integer val‐
441 ues to mean ``days''. The semantics of this was quite loose given that
442 find(1) offers different precision and semantics across platforms.
443 The current implementation of pmlogger_daily uses find-filter(1) which
445 provides high precision intervals and semantics that are relative to
446 the time of execution and are consistent across platforms.
447
449 egrep(1), find-filter(1), PCPIntro(1), pmconfig(1), pmlc(1), pmlog‐
450 conf(1), pmlogctl(1), pmlogextract(1), pmlogger(1), pmlogger_check(1),
451 pmlogger_daily_report(1), pmlogger_merge(1), pmlogmv(1), pmlo‐
452 grewrite(1), systemd(1), xz(1) and cron(8).
453Performance Co-Pilot PCP PMLOGGER_DAILY(1)