view release on metacpan or  search on metacpan

doc/recs-toptable.pod  view on Meta::CPAN

 {"priority":0,"pid":2,"rss":0,"state":"sleep"}
 {"priority":0,"pid":3,"rss":0,"state":"sleep"}
 {"priority":0,"pid":4,"rss":0,"state":"sleep"}
 {"priority":19,"pid":5,"rss":0,"state":"sleep"}
 {"priority":19,"pid":2094,"rss":8351744,"state":"run"}
 {"priority":0,"pid":28129,"rss":4784128,"state":"run"}
 {"priority":19,"pid":28171,"rss":405504,"state":"run"}
 
 Ok, Now lets get a table out of this, first we'll collate into some useful
 information:
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count
 {"priority":0,"count":4,"state":"sleep"}
 {"priority":19,"count":1,"state":"sleep"}
 {"priority":0,"count":1,"state":"run"}
 {"priority":19,"count":2,"state":"run"}
 
 And lets get a table:
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count | recs-toptable --x priority --y state
 +-----+--------+-+--+
 |     |priority|0|19|
 +-----+--------+-+--+
 |state|        | |  |
 +-----+--------+-+--+
 |run  |        |1|2 |
 +-----+--------+-+--+
 |sleep|        |4|1 |
 +-----+--------+-+--+
 
 So, you can see that the VALUES of priority and state are used as the columns /
 rows. So that there is 1 process in state 'run' and priority 0, and 4 in state
 'sleep' and priority 0
 
 The --cube option on recs-collate also interacts very well with toptable:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count --cube | recs-toptable --x priority --y state
 +-----+--------+-+--+---+
 |     |priority|0|19|ALL|
 +-----+--------+-+--+---+
 |state|        | |  |   |
 +-----+--------+-+--+---+
 |ALL  |        |5|3 |8  |
 +-----+--------+-+--+---+
 |run  |        |1|2 |3  |
 +-----+--------+-+--+---+
 |sleep|        |4|1 |5  |
 +-----+--------+-+--+---+
 
 We added an ALL row and an ALL column. So from this you can see that there are 5
 processes in priority 0, 3 processes in state 'run' and 8 processes all told in
 the table (the ALL, ALL intersection)
 
 Now lets see what happens when we have more than 1 left over field. Lets also
 sum up the rss usage of the processes with -a sum,rss on recs-collate:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count --cube -a sum,rss
 {"priority":0,"count":4,"state":"sleep","sum_rss":471040}
 {"priority":"ALL","count":5,"state":"sleep","sum_rss":471040}
 {"priority":19,"count":1,"state":"sleep","sum_rss":0}
 {"priority":0,"count":5,"state":"ALL","sum_rss":5255168}
 {"priority":0,"count":1,"state":"run","sum_rss":4784128}
 {"priority":"ALL","count":8,"state":"ALL","sum_rss":14012416}
 {"priority":"ALL","count":3,"state":"run","sum_rss":13541376}
 {"priority":19,"count":3,"state":"ALL","sum_rss":8757248}
 {"priority":19,"count":2,"state":"run","sum_rss":8757248}
 
 So now we have 2 left over fields that aren't columns, count and sum_rss. What
 happens to our table now:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count --cube -a sum,rss | recs-toptable --x priority --y state
 +-----+--------+-------+-------+--------+
 |     |priority|0      |19     |ALL     |
 +-----+--------+-------+-------+--------+
 |state|        |       |       |        |
 +-----+--------+-------+-------+--------+
 |ALL  |        |5255168|8757248|14012416|
 +-----+--------+-------+-------+--------+
 |run  |        |4784128|8757248|13541376|
 +-----+--------+-------+-------+--------+
 |sleep|        |471040 |0      |471040  |
 +-----+--------+-------+-------+--------+
 
 We now have sum_rss values in this field. What if we want the other field
 (count) displayed? We just use --v-field to specify what value field to use:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count --cube -a sum,rss | recs-toptable --x priority --y state --v count
 +-----+--------+-+--+---+
 |     |priority|0|19|ALL|
 +-----+--------+-+--+---+
 |state|        | |  |   |
 +-----+--------+-+--+---+
 |ALL  |        |5|3 |8  |
 +-----+--------+-+--+---+
 |run  |        |1|2 |3  |
 +-----+--------+-+--+---+
 |sleep|        |4|1 |5  |
 +-----+--------+-+--+---+
 
 Ok, but what if we want to see both left over fields at the same time? What we
 really want is to add a column or row for each of count and sum_rss. (where the
 title of the row is count or sum_rss, not the values of the field). We can do
 this by using the special FIELD specifier like so:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --key priority,state -a count --cube -a sum,rss | recs-toptable --x priority,FIELD --y state
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |     |priority|0    |       |19   |       |ALL  |        |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |     |FIELD   |count|sum_rss|count|sum_rss|count|sum_rss |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |state|        |     |       |     |       |     |        |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |ALL  |        |5    |5255168|3    |8757248|8    |14012416|
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |run  |        |1    |4784128|2    |8757248|3    |13541376|
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |sleep|        |4    |471040 |1    |0      |5    |471040  |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 
 So, now in one table we can see all the intersections of state and priority
 values with the count and sum_rss fields. Remember that the ALL field (row and
 column) are provided by the --cube functionality of recs-collate
 
 Now, say you want to pin value, lets just look at processes in state run
 for instance:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --cube --key priority,state -a count -a sum,rss | recs-toptable --x priority,FIELD --y state -v sum_rss,count --pin state=run
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |     |priority|0    |       |19   |       |ALL  |        |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |     |FIELD   |count|sum_rss|count|sum_rss|count|sum_rss |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |state|        |     |       |     |       |     |        |
 +-----+--------+-----+-------+-----+-------+-----+--------+
 |run  |        |1    |4784128|2    |8757248|3    |13541376|
 +-----+--------+-----+-------+-----+-------+-----+--------+
 
 As you can see, this is basically short hand for doing a recs-grep, the
 transformation to recs group would look like:
 
 $ cat /var/tmp/psrecs | recs-collate --perfect --cube --key priority,state -a count -a sum,rss | recs-grep '$r->{state} eq "run"' | recs-toptable --x priority,FIELD --y state -v sum_rss,count
 
 (which produces the same table as above).
 
 Help from: --help-keygroups:
 KEY GROUPS
    SYNTAX: !regex!opt1!opt2... Key groups are a way of specifying multiple
    fields to a recs command with a single argument or function. They are
    generally regexes, and have several options to control what fields they
    match. By default you give a regex, and it will be matched against all first
    level keys of a record to come up with the record list. For instance, in a
    record like this:
 
    { 'zip': 1, 'zap': 2, 'foo': { 'bar': 3 } }
 
    Key group: !z! would get the keys 'zip' and 'zap'
 
    You can have a literal '!' in your regex, just escape it with a \.
 
    Normally, key groups will only match keys whose values are scalars. This can
    be changed with the 'returnrefs' or rr flag.
 
    With the above record !f! would match no fields, but !f!rr would match foo
    (which has a value of a hash ref)
 
    Options on KeyGroups:
       returnrefs, rr  - Return keys that have reference values (default:off)
       full, f         - Regex should match against full keys (recurse fully)
       depth=NUM,d=NUM - Only match keys at NUM depth (regex will match against
                         full keyspec)
       sort, s         - sort keyspecs lexically
 
 Help from: --help-keyspecs: