forked from xrmx/bootchart
-
Notifications
You must be signed in to change notification settings - Fork 0
/
TODO
81 lines (57 loc) · 2.48 KB
/
TODO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
** Bugs:
* FIXME: are we merging / glupping threads properly ?
+ how can sreadahead apparently be doing no I/O ? :-)
+ extend rectangles in rendered output to note when
processes truly exit to improve duration rendering.
+ expand the tasks API for this.
* Tracing metrics:
+ traced %d processes of $$ (!) :-)
+ consumed %d seconds of CPU [ render these ... ]
+ horizontal zoom - should keep view horiz centered:
but does not - cf. set_xscale
+ merge processes for CPU counts:
+ if the ns count is -identical- (and non-zero) it is the same process
+ need to back-merge somehow ?
+ 'forked' ? ...
+ or can we do this by pid ? [ prolly ! ]
+ fix cumulative graphing - and boundary box bits
+ re-write ever more of bootchartd from shell into C ...
+ make the --dump thing return a success status,
+ poll the remote process (with ptrace ? - hmm ...)
+ to see if it has finished (?)
+ a good idea ? or will it introduce error ?
+ hmm.
+ have a 20 second delay before starting to poll it ?
** Features:
Add some tabs to the interactive renderer to show the raw files:
kernel dmesg, header, and perhaps the full (processed) cmdline2 log
Use the same stats that Intel's tracer uses of I/O / scheduling foo ...
+ use them to avoid taskstats polling if there has been
no scheduled timeslice (?)
Graph the swap / I/O delay stuff inside the bar (?)
Account for the time taken to run the bootchart-collector ...
+ scale all other times by this ammount [!] ;-)
+ to make it -appear- as if it doesn't impact the system :-)
* render I/O data backwards ... :-)
+ back-propagate delays that are longer than a rendered slice
Scale time discontinuously ? [ make the first seconds much longer ? ]
+ fun ! :-)
Remove 'merge_logger' - no longer necessary - it's just one app ...
+ potentially we should hide / remove it though :-)
* Better data processing / shrinkage:
+ post process in C code as we dump the data
+ why process the data in python ? can't we do it earlier ?
+ adjust the previous sample etc.
+ /proc/stat post-processing:
+ which fields do we actually parse & need ?
Process taskstats in advance
+ elide all non-useful lines:
+ ie. no-changes in various fields
* Export ODF, with a number of pie charts in it etc.
+ export cumulative data for rendering in ods ?
+ rather than adding manual rendering - which sucks (!?)
+ fun ... :-)
Notes:
+ SUSE initrd debugging; use "'shell=1'" - is dead good ...
#18a303 - Tango / Green
#2da31b - to same saturation