|
|
ab74e27bb8
|
Made the diff stat from inside the testfs, so new users/groups would stat correctly. Changed disco-ball fetch to fetch into the testfs, not into / (yikes!). New disco-ball command spin does what the main disco loop used to do for each module. Added the "users" disco ball to the standard library.
|
2012-08-23 00:16:22 -04:00 |
|
|
|
4c6d422395
|
Added builtin performance metrics. Add REPORT=true to one of your disco dance runs to print the report at the end, or use 'disco report' to produce a report on the last disco dance; reports are cleared at the beginning of every disco dance.
_internal diff is for the time taken to produce the filesystem diff
_internal fetch_params is for how long it took to fetch all the client parameters
Currently these performance numbers only account for about 50-70% of disco's activity, not sure how to report on the rest yet.
Example output:
report: _internal: diff
report: time_real 1.15 : time_user 0.21 : time_sys 0.92
report: mem_avg 0 : mem_max 4816 : mem_faults_major 0 : mem_faults_minor 25400
report: io_fsin 0 : io_fsout 16 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: _internal: fetch_params
report: time_real 1.14 : time_user 0.03 : time_sys 0.07
report: mem_avg 0 : mem_max 11152 : mem_faults_major 0 : mem_faults_minor 1674
report: io_fsin 0 : io_fsout 0 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: othermodule-3.2: fetch
report: time_real 2.27 : time_user 0.06 : time_sys 0.13
report: mem_avg 0 : mem_max 11136 : mem_faults_major 0 : mem_faults_minor 2842
report: io_fsin 0 : io_fsout 24 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: othermodule-3.2: template
report: etc/othermodule/stuff.cfg :
report: time_real 0.53 : time_user 0.03 : time_sys 0.35
report: mem_avg 0 : mem_max 4656 : mem_faults_major 119 : mem_faults_minor 5086
report: io_fsin 25408 : io_fsout 16 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: testmodule-1.0: exec
report: 00-hello.sh :
report: time_real 0.29 : time_user 0.03 : time_sys 0.20
report: mem_avg 0 : mem_max 4592 : mem_faults_major 54 : mem_faults_minor 3871
report: io_fsin 10992 : io_fsout 8 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: 10-service_stop.sh :
report: time_real 0.46 : time_user 0.03 : time_sys 0.32
report: mem_avg 0 : mem_max 4592 : mem_faults_major 110 : mem_faults_minor 4564
report: io_fsin 22328 : io_fsout 8 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
report: testmodule-1.0: fetch
report: time_real 2.24 : time_user 0.05 : time_sys 0.13
report: mem_avg 0 : mem_max 11152 : mem_faults_major 0 : mem_faults_minor 2894
report: io_fsin 0 : io_fsout 24 : io_sockin 0 : io_sockout 0 : io_signals 0
report: exit: 0
|
2012-08-17 20:46:57 -04:00 |
|
|
|
3cfffd192e
|
Moved disco colorization into a function, can be used standalone with "client/bin/disco colorize" on stdin.
Fixed disco-fs-diff to only use rsync, sed, grep, etc to produce the filesystem changelog, python script is gone
Minor fixups to other scripts
|
2012-08-17 19:24:25 -04:00 |
|
|
|
7d6d94ae3d
|
Moved disco-ball to universe, added functions necessary for fetching, templating, and executing a given disco ball. Made all existing scripts aware of NOOP and how to change their DISCOROOT accordingly. Added skeleton restricted.d/* files for a semi-safe base system. Added disco-param that allows management of parameters on the client (currently isn't smart enough to manage them on the server, or per-module). Added client/bin/disco that actually allows the entire thing to come together and get executed. Still lots of bugs to work out.
|
2012-08-08 05:30:03 -04:00 |
|
