summaryrefslogtreecommitdiffstats
path: root/TODO.virt-top
blob: 186aa42a04e9dcd08b6ff71f998cc7eb52438ac3 (plain)
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
82
83
84
85
86
87
88
89
90
91
92
93
This is a list of bugs & tasks for virt-top.
----------------------------------------------------------------------

*** 0 Use public CVS and bug tracker to replace this file (!)

*** 1 Segfault [FIXED 0.3.2.6]

I have seen this segfault once:
https://www.redhat.com/archives/libvir-list/2007-August/msg00214.html

It appeared to happen when several domains were shut down at the same
time, but has since been unreproducible.  Suspect crazy reference
counting in libvirt to be part of the problem.

[As suspected, turned out to be the crazy reference counting in
libvirt].

*** 2 Virt-top fails silently when started as non-root [FIXED 0.3.2.5]

*** 3 Irix/Solaris mode

'I' (toggles between 100% = all CPUs / 100% = single CPU).
In top this is called "Irix/Solaris mode" :-)

*** 4 [jmh] Build out the stats collected by the --csv <filename.csv> option.
    [DONE 0.3.2.9]

Suggested being able to give a selection of VMs to monitor from the
command line.  Also, being able to monitor just particular subsystems
(-sc for CPU, -sm for memory, etc. as in collectl).

*** 5 [jmh & jneedham] Monitor resources used by dom0 on behalf of the guests.

So that if dom0 is throttled right down, is that affecting throughput
of guests?  I honestly don't know if it's possible to do this, but
I'll raise the issue upstream to see if they have any suggestions.

*** 6 Per-guest / per-interface network statistics. [PARTIALLY DONE 0.3.2.6]

Have a separate screen which breaks out the domain name / device name
and gives complete interface stats.

RXBY TXBY RXRQ TXRQ DOMAIN   INTERFACE
  5M    0   48    0 debian   vif2.0
  2M    0   20    0 oracle   vif3.1
  1M   5K    8    2 oracle   vif3.0 

*** 7 Per-guest / per-device block statistics. [PARTIALLY DONE 0.3.2.6]

Have a separate screen which breaks out the domain name / device name
and gives complete block stats.

RDBY WRBY RDRQ WRRQ DOMAIN   DEVICE
  5M    0   48    0 debian   hdb
  2M    0   20    0 debian   hda
  1M   5K    8    2 oracle   hda

*** 8 [jneedham asked for] %WAIT

I'm assuming this would mean in the context of the hypervisor, how
long are runnable domains waiting in the runqueue for access to a free
pCPU.  I will ask upstream whether this stat is available through Xen,
and if so put in place the infrastructure to monitor it.

*** 9 [jmh asked for] Configuration file or environment variable [DONE 0.3.2.8]

Thinking about maybe this is something we could controly via a
.virt-top-init file or an environment variable ?  The same could also
be true for other options such as the selection of subsystems and/or
ordering/sorting ?

*** 10 [jmh] End time for data collection [DONE 0.3.3.0]

In --script mode allow an end time to be specified (when the program
quits).  Be nice to allow the end time to also be specified relative
to the current time.

*** 11 [jmh] Roll over CSV log file
    [PUNTED, but documented a way to do it using ordinary shell
     scripting in the manual page]

In --csv mode, allow the CSV file to be rolled over and compressed
periodically (eg. every so often, specific time of day, etc.)

*** 12 [cywse] How much memory is a guest actually using

Can we look inside the guest to find out how much memory the guest
is actually using?  Scenario is: One guest is maxed out and swapping
while other guests aren't using much of their allocated memory.

*** 13 [jmh] Display reason why disk & network stats cannot be displayed.

*** 14 [jmh] Show version of libvirt compiled against, running against.