I deployed an app to a Linode 360 VPS, and have notices a huge
fluctuation in performance, ie 50req/sec down to 2req/sec. Seeing as it
was at 50 at one time I can’t see the reason being in the code.
I just wanted to find out if you people have had experiences like this
and may have a solution. I don’t really know where to start in find a
solution for this, is it something I can fix, or is it the VPS server?
Thanks
How much memory is allocated to your VPS? Are the same controllers
fluctuating, or are they different controllers?
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
-I have 360MB of mem with a 512MB swap.
-All the requests that I am testing that return the huge variance is on
the same controller action.
-I don’t have an caching mainly because it is not a site for a large
number of people.
-I am using Debian 4.0 as an OS
If that will shed any more light on things
How man Mongrels are you running within that config?
Are you monitoring them with Monit?
What’s the average memory consumption of mongrel in your app?
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
On Jan 7, 2008, at 8:43 PM, Chris O. wrote:
Thanks-’
As far as I know linode uses User Mode Linux for it’s VPS’s. User
mode linux runs in userspace so VM’s running under this system can be
affected by the ‘bad neighbor’ affect a lot easier then say a Xen
based VPS. So someone else on the same box is probably sucking up the
resources when you see slow times.
Either that or you may just be swapping a lot. 360Mb is not a lot for
a rails app vps and if you are hitting swap it will kill performance
badly.
Cheers-
Where is a good recommended host? I know Ezra’s stuff is probably a
little too expensive for me :), but in couple months I will hopefully
be looking for something that can handle a decent load.
I always hear about slice host, but I figured that Linode offered the
same, or so I thought, and it was cheaper, so it won out for me.
I need some place that if the situation arises that upgrading will not
be an issue, no matter how much the upgrade is.
Thanks for the advice guys.
Chris,
Email me off list and I can get you a quote. Also check out
RailsMachine.com.
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
Jesse P. wrote:
How man Mongrels are you running within that config?
Are you monitoring them with Monit?
What’s the average memory consumption of mongrel in your app?
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
3 mongrels
I just have a tails -f on ssh right now. I will check out Monit though.
Here is my mem…I admit I am a little short on it.
total used free shared
buffers cached
Mem: 356008 339348 16660 0 40568
135320
-/+ buffers/cache: 163460 192548
Swap: 524280 0 524280
I will definately keep a closer eye on the correlation of the memory
available and performance.
I don’t know why the memory would vary that much. Right now the site has
just been released, so it is only me that is hitting it.
Jesse P. wrote:
Chris,
Email me off list and I can get you a quote. Also check out
RailsMachine.com.
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
You wouldn’t happen to have an email address would you? 
[email protected]
Thanks,
Jesse P.
Blue Box Group, LLC
p. 800-613-4305 x 801
I would highly recommend BlueBoxGroup. We went from SliceHost to them
and
are using a Big5 load balancer that is making our lifes so much easier.
AEM
On Jan 7, 2008 10:13 PM, Chris O. [email protected]
wrote:
Thanks for the advice guys.
Posted via http://www.ruby-forum.com/.
–
Adrian Esteban Madrid
Lead Developer, Prefab Markets
http://www.prefabmarkets.com