[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: statslog perl filter

At 08:01 AM 10/27/2004, Hallvard B Furuseth wrote:
>I've found this script useful when playing with server logs.
>Should I commit it somewhere?


>It doesn't really fit into the current
>OpenLDAP source tree, but I don't know where else to put it - there is
>no 'OpenLDAP tools' distribution.  Maybe we should start one.
>   http://folk.uio.no/hbf/OpenLDAP/statslog
>Usage message:
>  Usage: statslog [options] [logfiles; may be .gz or .bz2 files]
>  Output selected parts of slapd's statslog output (loglevel 256) from
>  syslog or stderr, grouping log lines by LDAP connection.
>  Options:
>    --brief       -b      Brief output (omit time, host/process name/ID).
>    --exclude=RE  -e RE   Exclude connections whose output matches REgexp.
>    --include=RE  -i RE   Only include connections matching REgexp.
>    --EXCLUDE=RE  -E RE   As --exclude, but case-sensitive.
>    --INCLUDE=RE  -I RE   As --include, but case-sensitive.
>    --loose       -l RE   Exclude "loose" lines (lines with no connection)
>                          matching RE.  E.g. -l'$' excludes all loose lines.
>    --sort        -s      Sort the input files, oldest first.
>    --trace       -t      Print file names as they are read.  Implies -u.
>    --units       -u      Print active connections at end of each file.
>  Multiple --exclude and --include options may be used.
>  Note that --trace and --units make the --exclude and --include options
>  unreliable for connections that span more than one log file.
>BTW, I'm a bit unsatisfied with the --loose option.  It sort of does the
>opposite of what it looks like - exclude instead of include.  But I've
>found it's current functionality the most convenient one, except perhaps
>'exclude all' would be a nicer default - and even less intuitive.
>Suggestions welcome.