[ANN] Radish, a monitoring and analysis service for Redis

[Nick Quaranto]

Hi everyone!

I'm happy to announce that Radish, a new monitoring and analysis
service for Redis, is finally publicly available. This was created out
of a serious need for introspection as to what Redis is doing, and of
course, many cries for help here on the mailing list about to get data
for a running instance.

Blog post:
http://bit.ly/igEiZ2

Sign up here:
http://radishapp.com

If you have any questions, feel free to ask!

-Nick

[Sergei Tulentsev]

Seems like a great service. However, it didn't work for me out of the box (see my comment under the blog post).

Do you have idea about the prices? Or maybe you can give some free accounts in exchange for a blog post :-)

--
You received this message because you are subscribed to the Google Groups "Redis DB" group.
To post to this group, send email to redi...@googlegroups.com.
To unsubscribe from this group, send email to redis-db+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/redis-db?hl=en.

--
Best regards,
Sergei Tulentsev

[Sergei Tulentsev]

It doesn't seem to catch pub/sub commands. I launched it against a redis instance that I use exclusively for pubsub, and I don't see anything in the dashboard.

On Tue, May 31, 2011 at 6:30 PM, Nick Quaranto <ni...@quaran.to> wrote:

--
You received this message because you are subscribed to the Google Groups "Redis DB" group.
To post to this group, send email to redi...@googlegroups.com.
To unsubscribe from this group, send email to redis-db+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/redis-db?hl=en.

[Dvir Volk]

Can it handle multiple redis instances on a single "project"? from a glimpse it seems like it can't.

Running quite a few instances as a single application, it would be nice to be able to get an overall view of them as one unit, and a deeper individual view.

Something like Ganglia's cluster view that merges CPU, memory, load etc from many machines to a single graph.

I did not use it yet, so forgive me if I've missed it and it does work like that.

[Nick Quaranto]

Hey Sergei,

Right now Radish counts PUBLISH and SUBSCRIBE as "other" commands.
I've tracked pub/sub with Radish before so I can vouch that it
*should* work! Are you connecting to the right instance (and the right
database?) Open up a support request on http://support.radishapp.com
and I'll get you sorted out.

Thanks,
Nick

[Nick Quaranto]

Hi Dvir,

Right now it treats each Redis instance as a separate entity. There's
no overall view *yet*, but you're not the first to ask for such an
"overview" look on your instances.

I think this is within our grasp, I definitely want to make Radish
work great for people that use single Redis instances and also
clusters of instances as well. Please fire off a suggestion for us at
http://support.radishapp.com so others can chime as well and we can
get a better sense of what to build.

Thanks!
Nick

[Sergei Tulentsev]

Yeah, it worked eventually. However, results were, mmm, unexpected. Memory usage and key count got their values from this second daemon (pub/sub one).

So I vote for grouping as well. Let's say, I have several groups of redises that are being used for different purposes. I don't want to mix them all together, but I do want to see some aggregate stats within each group. And probably a grand total (ops/sec, memory)

[Dean]

Hi Nick,

Radish looks like a great service! Unfortunately, we're already
running a centralized monitoring solution that we're pretty happy
with.

I noticed that you are trending metrics like ops/sec and
<command_type>/sec, which I don't believe are available in the INFO
page. Would you mind sharing some details as to how you obtain those
statistics from Redis? Is this done using MONITOR? If so, what are
the performance implications of using it on an ongoing basis?

Thanks,

Dean

On May 31, 6:08 pm, Nick Quaranto <n...@quaran.to> wrote:
> Hi Dvir,
>
> Right now it treats each Redis instance as a separate entity. There's
> no overall view *yet*, but you're not the first to ask for such an
> "overview" look on your instances.
>
> I think this is within our grasp, I definitely want to make Radish
> work great for people that use single Redis instances and also

> clusters of instances as well. Please fire off a suggestion for us athttp://support.radishapp.comso others can chime as well and we can

> get a better sense of what to build.
>
> Thanks!
> Nick
>
>
>
>
>
>
>
> On Tue, May 31, 2011 at 6:44 PM, Dvir Volk <dvir...@gmail.com> wrote:
> > Can it handle multiple redis instances on a single "project"? from a glimpse
> > it seems like it can't.
> > Running quite a few instances as a single application, it would be nice to
> > be able to get an overall view of them as one unit, and a deeper individual
> > view.
> > Something like Ganglia's cluster view that merges CPU, memory, load etc from
> > many machines to a single graph.
> > I did not use it yet, so forgive me if I've missed it and it does work like
> > that.
>

[Nick Quaranto]

Yes, it's all done using MONITOR and a lot of data munging. Daikon,
the Radish client daemon, is open-source if you want to read it:
http://github.com/thoughtbot/daikon.

We've seen performance hits with *very* high traffic instances (we're
talking 15k+ ops/sec, which is huge for a webapp) but mostly it hasn't
been a problem. We've already done some performance optimization on
Daikon and will continue to as we grow out.