vSphere / Vcenter exporter

3,422 views
Skip to first unread message

manda.u...@gmail.com

unread,
Apr 21, 2017, 2:19:10 PM4/21/17
to Prometheus Users
is there any vSphere or VCenter exporter available to pull the VMWare stats and events into Promethius ?

Toni Schmidbauer

unread,
Apr 21, 2017, 3:47:23 PM4/21/17
to manda.u...@gmail.com, Prometheus Users
manda.u...@gmail.com writes:
> is there any vSphere or VCenter exporter available to pull the VMWare
> stats and events into Promethius ?

afaik not an official one. but a quick google found this:

https://github.com/brandonweeks/vsphere_exporter

seems unfinished, but worth a try. we are also very interested in a
vsphere exporter. i'll try to get the one listed above working within
the next weeks. my go knowledge is limited so this could take a while
:-)

hth
toni

Mike Knell

unread,
Apr 24, 2017, 3:49:44 AM4/24/17
to Toni Schmidbauer, manda.u...@gmail.com, Prometheus Users
FWIW, I’ve been working (slowly, as I had to learn go in the meantime) on getting some useful metrics out of VSphere over the last couple of weeks - the existing exporter is pretty crude and low-level, and we’re more interested in monitoring at the VM level. Happy to share the work in progress, although I think “release” would be a slightly excessive term for it.

Mike

Toni Schmidbauer

unread,
Apr 24, 2017, 5:31:08 AM4/24/17
to Mike Knell, Toni Schmidbauer, manda.u...@gmail.com, Prometheus Users
Mike Knell <m...@uffish.net> writes:
> FWIW, I’ve been working (slowly, as I had to learn go in the meantime)
> on getting some useful metrics out of VSphere over the last couple of
> weeks - the existing exporter is pretty crude and low-level, and we’re
> more interested in monitoring at the VM level. Happy to share the work
> in progress, although I think “release” would be a slightly excessive
> term for it.

i would be really grateful if you could share your current work. don't
worry about quality, i'm a go newbie myself, so my guess is we will
learn both from the experience :-)

we are monitoring the vm's as well, but it would be nice to have some
basic performance data from the underlying esx hosts.

thanks
toni

Rémi Verchère

unread,
May 9, 2017, 12:08:52 PM5/9/17
to Prometheus Users, m...@uffish.net, toni+prome...@stderr.at, manda.u...@gmail.com, to...@stderr.at
Hi,


Le lundi 24 avril 2017 11:31:08 UTC+2, Toni Schmidbauer a écrit :
i would be really grateful if you could share your current work. don't
worry about quality, i'm a go newbie myself, so my guess is we will
learn both from the experience :-)

I've also started writing some VMWare exporter, you can check current work here:
https://github.com/rverchere/vmware_exporter

It only get VM Snapshots infos for now, I hope getting more metrics soon. Any help or suggestions are welcome.

Rémi 

Julius Volz

unread,
May 9, 2017, 2:20:53 PM5/9/17
to Rémi Verchère, Prometheus Users, m...@uffish.net, toni+prome...@stderr.at, manda.u...@gmail.com, to...@stderr.at
On Tue, May 9, 2017 at 6:08 PM, Rémi Verchère <re...@verchere.fr> wrote:
Hi,

Le lundi 24 avril 2017 11:31:08 UTC+2, Toni Schmidbauer a écrit :
i would be really grateful if you could share your current work. don't
worry about quality, i'm a go newbie myself, so my guess is we will
learn both from the experience :-)

I've also started writing some VMWare exporter, you can check current work here:
https://github.com/rverchere/vmware_exporter

Short feedback on the metric naming. By our conventions, it would be better to change:

- Change "vmware_snapshot_count" to "vmware_snapshots" ("_count" is reserved for counters resulting from histograms / summaries). I assume this is the current number of existing snapshots, not a total counter over time (since the metric type is a gauge). Expanding on the HELP text a bit could help.

- "vmware_snapshot_age" is missing the unit in the metric name (which should always be seconds, not days), but it would also be better to work with if it was a Unix timestamp of the last snapshot instead an age: "vmware_snapshot_timestamp_seconds". Then you can formulate alerts like:

  time() - vmware_snapshot_timestamp_seconds > x

Julius Volz

unread,
May 9, 2017, 2:21:28 PM5/9/17
to Rémi Verchère, Prometheus Users, m...@uffish.net, toni+prome...@stderr.at, manda.u...@gmail.com, to...@stderr.at

Rémi Verchère

unread,
May 9, 2017, 4:34:48 PM5/9/17
to Julius Volz, Prometheus Users
Short feedback on the metric naming. By our conventions, it would be better to change:
[...]

Ok thanks I'll change it, and will read deeper the Metric and label naming page for next metric names.

sateesh...@gmail.com

unread,
Jan 28, 2018, 3:28:35 PM1/28/18
to Prometheus Users
I'm very new to this stuff - excuse me to ask basic questions

I'm core VMware Administrator and need to enable performance charts for Storage IOPs via Grafana/Prometheus 

I saw the code available but do I just need to execute the scripts?  I installed Ubuntu VM and Promethues on it - now I just need to copy the files from your vsphere-exporter?  

Any inputs are highly appreciated.

Thanks
T Sateesh
Reply all
Reply to author
Forward
0 new messages