xen usage monitoring (munin plugin)

classic Classic list List threaded Threaded
17 messages Options
Reply | Threaded
Open this post in threaded view
|

xen usage monitoring (munin plugin)

Goetz Bock
Dear list,

as my Xen3 server is finaly becomming ready for production I was
wondering if anyone has written a xen usage monitoring/logging tool?

Something like xm top, but with continuose logging to a (rr)database.

Or a munin plugin?

I can't remember seeing anything simmilar on the list.

BTW: where does xm top get the NETTX(k) and NETRX(k) from?

For a munin plugin I'd imagin two modes:

- just monitor cpu usage off all domains (dom0 and domU) in one graph.
  This can be parsed from xm list, or as suggested 2004/12/xx using some
  more low level xen api. But I don't know how to figure out which of
  the real CPU was used, otoh this information might not be relevant at
  all.
- monitor just one domain. But all data, like xm top does.
--
/"\ Goetz Bock at blacknet dot de  --  secure mobile Linux everNETting
\ /       (c) 2006 Creative Commons, Attribution-ShareAlike 2.0 de
 X   [ 1. Use descriptive subjects - 2. Edit a reply for brevity -  ]
/ \  [ 3. Reply to the list - 4. Read the archive *before* you post ]

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

tatu
On Sat, 7 Jan 2006, Goetz Bock wrote:

> Dear list,
>
> as my Xen3 server is finaly becomming ready for production I was
> wondering if anyone has written a xen usage monitoring/logging tool?

funny you should ask, I wrote one yesterday :)

> Something like xm top, but with continuose logging to a (rr)database.
>
> Or a munin plugin?

its a munin plugin, but it doesnt work right just yet, it just gets times
from xm list (and names of course)

> I can't remember seeing anything simmilar on the list.

yeah googled around for some time, but couldnt find any

> BTW: where does xm top get the NETTX(k) and NETRX(k) from?
>
> For a munin plugin I'd imagin two modes:
>
> - just monitor cpu usage off all domains (dom0 and domU) in one graph.
>  This can be parsed from xm list, or as suggested 2004/12/xx using some
>  more low level xen api. But I don't know how to figure out which of
>  the real CPU was used, otoh this information might not be relevant at
>  all.
> - monitor just one domain. But all data, like xm top does.
Idont have xm top on this server so I cant write a plugin for it :/

by no means this plugin is complete and I dont take any responsibility
what ever and so on ;)

it should be quite easy to read...

--
Tatu Wikman
[hidden email]
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
In reply to this post by Goetz Bock
I would be interested in something like that for sure myself, not sure with munin as i haven't used it till now.

You can get the TX and RX bytes from ifconfig on the dom0.

On 1/8/06, Goetz Bock <[hidden email]> wrote:
Dear list,

as my Xen3 server is finaly becomming ready for production I was
wondering if anyone has written a xen usage monitoring/logging tool?

Something like xm top, but with continuose logging to a (rr)database.

Or a munin plugin?

I can't remember seeing anything simmilar on the list.

BTW: where does xm top get the NETTX(k) and NETRX(k) from?

For a munin plugin I'd imagin two modes:

- just monitor cpu usage off all domains (dom0 and domU) in one graph.
  This can be parsed from xm list, or as suggested 2004/12/xx using some
  more low level xen api. But I don't know how to figure out which of
  the real CPU was used, otoh this information might not be relevant at
  all.
- monitor just one domain. But all data, like xm top does.
--
/"\ Goetz Bock at blacknet dot de  --  secure mobile Linux everNETting
\ /       (c) 2006 Creative Commons, Attribution-ShareAlike 2.0 de
X   [ 1. Use descriptive subjects - 2. Edit a reply for brevity -  ]
/ \  [ 3. Reply to the list - 4. Read the archive *before* you post ]

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users



--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Goetz Bock
In reply to this post by tatu

On Sat, Jan 07 '06 at 21:44, [hidden email] wrote:
> >as my Xen3 server is finaly becomming ready for production I was
> >wondering if anyone has written a xen usage monitoring/logging tool?
>
> funny you should ask, I wrote one yesterday :)

I have written a simmilar script, in sh not in perl:

------------------------------------------------------------------------
#!/bin/sh
#
# Script to monitor CPU usage of Xen domains
#
# Parameters understood:
#
#     conifg    (required)
#     autoconf  (optinal - used by munin-config)
#

MAXDOMAINS=16

if [ "$1" = "autoconf" ]; then
        if which xm > /dev/null ; then
                echo yes
                exit 0
        fi
        echo "no (xm not found)"
        exit 1
fi

if [ "$1" = "config" ]; then

        echo 'graph_title Xen Domain Utilerisation'
        echo 'graph_args --base 1000 -l 0'
        echo 'graph_scale no'
        echo 'graph_vlable mS'
        echo 'graph_category xen'
        echo 'graph_info This graph shows of many mS wall time where used by a d
omain'
        xm list | grep -v "^Name .* Console$" | \
        while read name domid mem cpu state time console; do
                name=`echo $name | sed -e"s/-/_/"`
                echo "$name.label $name"
                echo "$name.type COUNTER"
#               if [ "$name" = "Domain_0" ]; then
#                       echo "$name.draw AREA"
#               else
#                       echo "$name.draw STACK"
#               fi
                echo "$name.min 0"
                echo "$name.info Wall clock time spend for $name"
        done
        exit 0
fi

        xm list | grep -v "^Name .* Console$" | \
        while read name domid mem cpu state time console; do
                name=`echo $name | sed -e"s/-/_/"`
                echo "$name.label $name"
                echo "$name.type COUNTER"
#               if [ "$name" = "Domain_0" ]; then
#                       echo "$name.draw AREA"
#               else
#                       echo "$name.draw STACK"
#               fi
                echo "$name.min 0"
                echo "$name.info Wall clock time spend for $name"
        done
        exit 0
fi

xm list | grep -v "^Name .* Console$" | \
while read name domid mem cpu state time console; do
        name=`echo $name | sed -e"s/-/_/"`
        time=`echo $time | sed -e "s/\.//"`
        echo "$name.value $time"
done
------------------------------------------------------------------------

But I'm not too proud of it. Unlike your version it:
- multiplies time by 10. So the comment is wrong, it's not mS but deca S
  or 10th of mS
- it should be able to handle adding/removing of domains

Still I would prefere it if it would log the actuall utilerisation, not
some "strange" CPU Time dirivat. Espcially as the values logged by my
plugin are in the 0.01 .. 0.22 range, with peeks at 1.72 (for this week)
or 5.19 (for the last year), and this is after multiplication by 10 as
done my the script.
--
Goetz Bock       (c) 2006 as     blacknet.de - Munich - Germany   /"\
IT Consultant  Creative Commons  secure mobile Linux everNETting  \ /
                                                                   X
 ASCII Ribbon Campaign against HTML email & microsoft attachments / \

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

John Levin
In reply to this post by tatu
On 1/7/06, [hidden email] <[hidden email]> wrote:

> On Sat, 7 Jan 2006, Goetz Bock wrote:
>
> > Dear list,
> >
> > as my Xen3 server is finaly becomming ready for production I was
> > wondering if anyone has written a xen usage monitoring/logging tool?
>
> funny you should ask, I wrote one yesterday :)
>
> > Something like xm top, but with continuose logging to a (rr)database.
> >
> > Or a munin plugin?
>
> its a munin plugin, but it doesnt work right just yet, it just gets times
> from xm list (and names of course)
>
> > I can't remember seeing anything simmilar on the list.
>
> yeah googled around for some time, but couldnt find any

What about XenMon?   It's included in the repo: tools/xenmon/

A recent post:
http://lists.xensource.com/archives/html/xen-users/2006-01/msg00134.html

>
> > BTW: where does xm top get the NETTX(k) and NETRX(k) from?
> >
> > For a munin plugin I'd imagin two modes:
> >
> > - just monitor cpu usage off all domains (dom0 and domU) in one graph.
> >  This can be parsed from xm list, or as suggested 2004/12/xx using some
> >  more low level xen api. But I don't know how to figure out which of
> >  the real CPU was used, otoh this information might not be relevant at
> >  all.
> > - monitor just one domain. But all data, like xm top does.
>
> Idont have xm top on this server so I cant write a plugin for it :/
>
> by no means this plugin is complete and I dont take any responsibility
> what ever and so on ;)
>
> it should be quite easy to read...
>
> --
> Tatu Wikman
> [hidden email]
>
> _______________________________________________
> Xen-users mailing list
> [hidden email]
> http://lists.xensource.com/xen-users
>
>
>

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
On 1/9/06, John Levin <[hidden email]> wrote:

What about XenMon?   It's included in the repo: tools/xenmon/

A recent post:
http://lists.xensource.com/archives/html/xen-users/2006-01/msg00134.html

Did you have any luck in running xenmon and understanding its output ? I followed that post enabled trace and after that when i run python xenmon.py, i get the following (even though my screen size is bigger than that)

xenbaked: no process killed
ms_per_sample = 100
Initialized with 1 cpu
CPU Frequency = 2014.82
Your terminal screen is not big enough; Please resize it.
row=2, col=90, str='Gotten'

By hit and trial i got it working, and this is what it shows when run

 CPU = 0        Last 10 seconds                              Last 1 second
 ==================================================================================================
  0   1.66 ms     0.17%       156.74 us/ex      17.15 ms    1.72%       196.03 us/ex      Gotten
  0                           7.41 ms/ex                                8.79 ms/ex        Allocated
  0   998.33 ms   99.83%      0.00 ns/io        982.72 ms   98.27%      0.00 ns/io        Blocked
  0   18.57 us    0.00%       1.76 us/ex        107.05 us   0.01%       1.22 us/ex        Waited
  0                           10/s                                      87                Execution
co0nt 0/s                     0/ex              0                       0.00/ex           I/O Count
  6   13.49 us    0.00%       147.75 us/ex      164.29 us   0.02%       85.43 us/ex       Gotten
  6                           500.00 us/ex                              500.00 us/ex      Allocated
  6   999.98 ms   100.00%     0.00 ns/io        999.81 ms   99.98%      0.00 ns/io        Blocked
  6   2.08 us     0.00%       22.83 us/ex       20.09 us    0.00%       10.45 us/ex       Waited
  6                           0/s                                       1                 Execution
co6nt 0/s                     0/ex              0                       0.00/ex           I/O Count
  31  110.47 ms   11.05%      10.45 ms/ex       982.59 ms   98.26%      11.35 ms/ex       Gotten
  31                          998.27 ms/ex                              1.00 s/ex         Allocated
  31  0.00 ns     0.00%       0.00 ns/io        0.00 ns     0.00%       0.00 ns/io        Blocked
  31  2.03 ms     0.20%       191.73 us/ex      17.31 ms    1.73%       200.01 us/ex      Waited
  31                          10/s                                      86                Execution
co31t 0/s                     0/ex              0                       0.00/ex           I/O Count
                  11.21%                                    99.99%

Can this output be redirected to something / parsed somehow  ?

--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Tim Freeman
In reply to this post by Goetz Bock
On Sat, 7 Jan 2006 20:33:51 +0100
Goetz Bock <[hidden email]> wrote:

> Dear list,
>
> as my Xen3 server is finaly becomming ready for production I was
> wondering if anyone has written a xen usage monitoring/logging tool?
>
> Something like xm top, but with continuose logging to a (rr)database.

There is this for CPU time -> rrd
http://skaya.enix.org/wiki/XenStats

I needed to change a tiny something to get it working in Xen3:

diff xenupdate.py modified.xenupdate.py
16c16
<     name,id,mem,cpu,state,cputime,console=re.split("[\t ]+",domain)
---
>     name,id,mem,cpu,state,cputime=re.split("[\t ]+",domain)


Tim



>
> Or a munin plugin?
>
> I can't remember seeing anything simmilar on the list.
>
> BTW: where does xm top get the NETTX(k) and NETRX(k) from?
>
> For a munin plugin I'd imagin two modes:
>
> - just monitor cpu usage off all domains (dom0 and domU) in one graph.
>   This can be parsed from xm list, or as suggested 2004/12/xx using some
>   more low level xen api. But I don't know how to figure out which of
>   the real CPU was used, otoh this information might not be relevant at
>   all.
> - monitor just one domain. But all data, like xm top does.
> --
> /"\ Goetz Bock at blacknet dot de  --  secure mobile Linux everNETting
> \ /       (c) 2006 Creative Commons, Attribution-ShareAlike 2.0 de
>  X   [ 1. Use descriptive subjects - 2. Edit a reply for brevity -  ]
> / \  [ 3. Reply to the list - 4. Read the archive *before* you post ]
>
> _______________________________________________
> Xen-users mailing list
> [hidden email]
> http://lists.xensource.com/xen-users
>

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
Thanks for the update on xenstats. xenupdate.py runs fine and creates the rrd files. However when running xengraph.py, it gives the following error:

--imgformat PNG --unit % --vertical-label CPU%  DEF:id0raw=/home/anand/root/xen-development/xen-graphs/Domain- 0.rrd:cpu:LAST CDEF:id0percent=id0raw,10,/ CDEF:id0zero=id0percent,DUP,UN,EXC,0,EXC,IF DEF:id1raw=/home/anand/root/xen-development/xen-graphs/vm01.rrd:cpu:LAST CDEF:id1percent=id1raw,10,/ CDEF:id1zero=id1percent,DUP,UN,EXC,0,EXC,IF AREA:id0zero#000000:Domain-0 GPRINT:id0percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j" STACK:id1zero#FF0000:vm01 GPRINT:id1percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j"
ERROR: unknown option '--unit'
--imgformat PNG --unit % --vertical-label CPU%  DEF:id0raw=/home/anand/root/xen-development/xen-graphs/Domain-0.rrd:cpu:LAST CDEF:id0percent=id0raw,10,/ CDEF:id0zero=id0percent,DUP,UN,EXC,0,EXC,IF DEF:id1raw=/home/anand/root/xen-development/xen-graphs/vm01.rrd:cpu:LAST CDEF:id1percent=id1raw,10,/ CDEF:id1zero=id1percent,DUP,UN,EXC,0,EXC,IF AREA:id0zero#000000:Domain-0 GPRINT:id0percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j" STACK:id1zero#FF0000:vm01 GPRINT:id1percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j"
ERROR: unknown option '--unit'
--imgformat PNG --unit % --vertical-label CPU%  DEF:id0raw=/home/anand/root/xen-development/xen-graphs/Domain-0.rrd:cpu:LAST CDEF:id0percent=id0raw,10,/ CDEF:id0zero=id0percent,DUP,UN,EXC,0,EXC,IF DEF:id1raw=/home/anand/root/xen-development/xen-graphs/vm01.rrd:cpu:LAST CDEF:id1percent=id1raw,10,/ CDEF:id1zero=id1percent,DUP,UN,EXC,0,EXC,IF AREA:id0zero#000000:Domain-0 GPRINT:id0percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j" STACK:id1zero#FF0000:vm01 GPRINT:id1percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j"
ERROR: unknown option '--unit'
--imgformat PNG --unit % --vertical-label CPU%  DEF:id0raw=/home/anand/root/xen-development/xen-graphs/Domain-0.rrd:cpu:LAST CDEF:id0percent=id0raw,10,/ CDEF:id0zero=id0percent,DUP,UN,EXC,0,EXC,IF DEF:id1raw=/home/anand/root/xen-development/xen-graphs/vm01.rrd:cpu:LAST CDEF:id1percent=id1raw,10,/ CDEF:id1zero=id1percent,DUP,UN,EXC,0,EXC,IF AREA:id0zero#000000:Domain-0 GPRINT:id0percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j" STACK:id1zero#FF0000:vm01 GPRINT:id1percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j"
ERROR: unknown option '--unit'
--imgformat PNG --unit % --vertical-label CPU%  DEF:id0raw=/home/anand/root/xen-development/xen-graphs/Domain-0.rrd:cpu:LAST CDEF:id0percent=id0raw,10,/ CDEF:id0zero=id0percent,DUP,UN,EXC,0,EXC,IF DEF:id1raw=/home/anand/root/xen-development/xen-graphs/vm01.rrd:cpu:LAST CDEF:id1percent=id1raw,10,/ CDEF:id1zero=id1percent,DUP,UN,EXC,0,EXC,IF AREA:id0zero#000000:Domain-0 GPRINT:id0percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j" STACK:id1zero#FF0000:vm01 GPRINT:id1percent:AVERAGE:"Average CPU usage\: %02.0lf%%\j"
ERROR: unknown option '--unit'

Its complaining about --unit. Any help would be appreciated.

On 1/10/06, Tim Freeman <[hidden email]> wrote:
On Sat, 7 Jan 2006 20:33:51 +0100
Goetz Bock < [hidden email]> wrote:

> Dear list,
>
> as my Xen3 server is finaly becomming ready for production I was
> wondering if anyone has written a xen usage monitoring/logging tool?
>
> Something like xm top, but with continuose logging to a (rr)database.

There is this for CPU time -> rrd
http://skaya.enix.org/wiki/XenStats

I needed to change a tiny something to get it working in Xen3:

diff xenupdate.py modified.xenupdate.py
16c16
<     name,id,mem,cpu,state,cputime,console=re.split ("[\t ]+",domain)
---
>     name,id,mem,cpu,state,cputime=re.split("[\t ]+",domain)


Tim



>
> Or a munin plugin?
>
> I can't remember seeing anything simmilar on the list.
>
> BTW: where does xm top get the NETTX(k) and NETRX(k) from?
>
> For a munin plugin I'd imagin two modes:
>
> - just monitor cpu usage off all domains (dom0 and domU) in one graph.
>   This can be parsed from xm list, or as suggested 2004/12/xx using some

>   more low level xen api. But I don't know how to figure out which of
>   the real CPU was used, otoh this information might not be relevant at
>   all.
> - monitor just one domain. But all data, like xm top does.
> --
> /"\ Goetz Bock at blacknet dot de  --  secure mobile Linux everNETting
> \ /       (c) 2006 Creative Commons, Attribution-ShareAlike 2.0 de
>  X   [ 1. Use descriptive subjects - 2. Edit a reply for brevity -  ]
> / \  [ 3. Reply to the list - 4. Read the archive *before* you post ]
>
> _______________________________________________
> Xen-users mailing list
> [hidden email]
> http://lists.xensource.com/xen-users
>

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users



--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Tim Freeman
On Tue, 10 Jan 2006 12:28:50 +0530
Anand <[hidden email]> wrote:

> Thanks for the update on xenstats. xenupdate.py runs fine and creates the
> rrd files. However when running xengraph.py, it gives the following error:
>
[....]
> ERROR: unknown option '--unit'
>
> Its complaining about --unit. Any help would be appreciated.


I ran into that as well, we're using a different version of rrdtool than the
author, find:

cmdline="--imgformat PNG --unit % --vertical-label CPU%  "

and remove "--unit %":

cmdline="--imgformat PNG --vertical-label CPU%  "


Tim



_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
Thanks it worked.

On 1/10/06, Tim Freeman <[hidden email]> wrote:
On Tue, 10 Jan 2006 12:28:50 +0530
Anand <[hidden email]> wrote:

> Thanks for the update on xenstats. xenupdate.py runs fine and creates the
> rrd files. However when running xengraph.py, it gives the following error:
>
[....]
> ERROR: unknown option '--unit'
>
> Its complaining about --unit. Any help would be appreciated.


I ran into that as well, we're using a different version of rrdtool than the
author, find:

cmdline="--imgformat PNG --unit % --vertical-label CPU%  "

and remove "--unit %":

cmdline="--imgformat PNG --vertical-label CPU%  "


Tim





--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
In reply to this post by Tim Freeman
I have the graphs being made perfectly fine now. However i don't seem to understand this. Can you look at this graph for me and tell me what does the left side scale 'm' stands for ?

Here is the output from xm list

Name                              ID Mem(MiB) VCPUs State  Time(s)
Domain-0                           0      256     1 r-----   277.7
vm01                               3       32     1 -b----     7.9

Thanks for the help.
--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users

xen-hourly[1].png (16K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Tim Freeman
On Tue, 10 Jan 2006 20:12:05 +0530
Anand <[hidden email]> wrote:

> I have the graphs being made perfectly fine now. However i don't seem to
> understand this. Can you look at this graph for me and tell me what does the
> left side scale 'm' stands for ?

Not, sure, ms?  Mine has CPU% from 0 -> 100

What version is your rrdtool?  You may need to look at that python code in
xengraph and see what the rrdtool invocation looks like and then type "rrdtool
graph" without any arguments to see what flags are available to you and try from
there. That change I sent works for RRDtool 1.2.6 here.

Tim


>
> Here is the output from xm list
>
> Name                              ID Mem(MiB) VCPUs State  Time(s)
> Domain-0                           0      256     1 r-----   277.7
> vm01                               3       32     1 -b----     7.9
>
> Thanks for the help.
> --
>
> regards,
>
> Anand
>

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
I am using rrd 1.2.12.

I read the python code. It has 2 cmdlines, one which keeps the left hand scale from 0->100 other doesn't. By default the 0->100 scale is disabled.

On 1/10/06, Tim Freeman <[hidden email]> wrote:
On Tue, 10 Jan 2006 20:12:05 +0530
Anand <[hidden email]> wrote:

> I have the graphs being made perfectly fine now. However i don't seem to
> understand this. Can you look at this graph for me and tell me what does the
> left side scale 'm' stands for ?

Not, sure, ms?  Mine has CPU% from 0 -> 100

What version is your rrdtool?  You may need to look at that python code in
xengraph and see what the rrdtool invocation looks like and then type "rrdtool
graph" without any arguments to see what flags are available to you and try from
there. That change I sent works for RRDtool 1.2.6 here.

Tim


>
> Here is the output from xm list
>
> Name                              ID Mem(MiB) VCPUs State  Time(s)
> Domain-0                           0      256     1 r-----   277.7
> vm01                               3       32     1 -b----     7.9
>
> Thanks for the help.
> --
>
> regards,
>
> Anand
>



--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Bugzilla from diwaker.lists@gmail.com
In reply to this post by Anand Gupta
Sorry I'm replying to this so late -- didn't see it earlier. More inline.

On 1/9/06, Anand <[hidden email]> wrote:

> Did you have any luck in running xenmon and understanding its output ? I
> followed that post enabled trace and after that when i run python xenmon.py,
> i get the following (even though my screen size is bigger than that)
>
> xenbaked: no process killed
> ms_per_sample = 100
> Initialized with 1 cpu
> CPU Frequency = 2014.82
> Your terminal screen is not big enough; Please resize it.
> row=2, col=90, str='Gotten'
>
>  By hit and trial i got it working, and this is what it shows when run
Doing 'xenmon.py -h' will give a lot of helpful information. In
particular, Xenmon can easily output the data to a file instead of the
online curses interface. Doing something like:

$ sudo xenmon.py -n -t 60 -p /tmp/foo

Will put xenmon in not-live mode (output to files), will run for 60
seconds, logging data at every 1 s (also configurable) and put the
data in /tmp/foo-dom0.log, /tmp/foo-domid.log and so on.

Hope that helps,
Diwaker
--
Web/Blog/Gallery: http://floatingsun.net

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
Thanks for the tip Diwaker. I guess who can explain better than the author himself ;)

Is there a way to store the data in files with domain name instead of its id ? Also what is the overhead which is caused by running xenmon in background to grab information related to domains ?

On 1/14/06, Diwaker Gupta <[hidden email]> wrote:
Sorry I'm replying to this so late -- didn't see it earlier. More inline.

On 1/9/06, Anand <[hidden email]> wrote:

> Did you have any luck in running xenmon and understanding its output ? I
> followed that post enabled trace and after that when i run python xenmon.py,
> i get the following (even though my screen size is bigger than that)
>
> xenbaked: no process killed
> ms_per_sample = 100
> Initialized with 1 cpu
> CPU Frequency = 2014.82
> Your terminal screen is not big enough; Please resize it.
> row=2, col=90, str='Gotten'
>
>  By hit and trial i got it working, and this is what it shows when run

Doing 'xenmon.py -h' will give a lot of helpful information. In
particular, Xenmon can easily output the data to a file instead of the
online curses interface. Doing something like:

$ sudo xenmon.py -n -t 60 -p /tmp/foo

Will put xenmon in not-live mode (output to files), will run for 60
seconds, logging data at every 1 s (also configurable) and put the
data in /tmp/foo-dom0.log, /tmp/foo-domid.log and so on.

Hope that helps,
Diwaker
--
Web/Blog/Gallery: http://floatingsun.net



--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Bugzilla from diwaker.lists@gmail.com
> Is there a way to store the data in files with domain name instead of its id
> ? Also what is the overhead which is caused by running xenmon in background
> to grab information related to domains ?

Currently xenmon uses domain IDs only, but it should be fairly easy to
extend it to use domain names instead. The frontend is just a python
script, so you could simply make your own copy and modify it as you
please.

I've stress tested xenmon under all kinds of settings and when writing
to log files the overhead is typically less than 1-2%.

Diwaker
--
Web/Blog/Gallery: http://floatingsun.net

_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users
Reply | Threaded
Open this post in threaded view
|

Re: xen usage monitoring (munin plugin)

Anand Gupta
On 1/14/06, Diwaker Gupta <[hidden email]> wrote:
Currently xenmon uses domain IDs only, but it should be fairly easy to
extend it to use domain names instead. The frontend is just a python
script, so you could simply make your own copy and modify it as you
please.

I've stress tested xenmon under all kinds of settings and when writing
to log files the overhead is typically less than 1-2%.

Is there a reporting tool which can take values from xenmon and perhaps keep qos delivered to various domains over a period of time in a db or something ?

Thanks.

--

regards,

Anand
_______________________________________________
Xen-users mailing list
[hidden email]
http://lists.xensource.com/xen-users