Bug ? ELB and replacing unhealthy servers

[Srinivasan Subramanian]

I use the ELB (instead of the nginx role) for load balancing.  When the instance is marked unhealthy on the ELB by AWS, Scalr is not detecting this and creating a new instance.  Instead ELB indicates its out of service and Scalr is just running the old instance.

 

Is this a bug or some setup issue in my landscape?  If the unhealthy instance is not detected and a new instance recreated, ELB would turn out to be useless right?

Cheers

Srini

[Srinivasan Subramanian]

I need to add this (critical) piece of info: My app runs on JBoss on a different port ie. 8080.  So ELB is set to check for health on 8080, whereas I guess Scalr is checking port 80?  Would that mean I need to custom script this?

 

Srini

[Srini]

Hello Scalr Team

Any hints / answers on this?  Depending on your answer I will need to work on a solution for this.  Appreciate your support

Cheers

Srini

[Srini]

Hi

Is anyone in the community using ELB with Scalr?  Have you come across the problem i detailed?  Maybe i am the only one facing it since Scalr monitors port 80 but my app runs on 8080 and ELB is also monitoring it on 8080.

Can some one from Scalr please confirm if this is supposed to work with my config or not?  I just need a confirmation .. based on the reply I will handle the refreshing of the instances somehow.

Thanks
Srini

[Sebastian Stadil]

Hey guys,

 

We've had an internal discussion on this, and I'd like to share our thoughts with you, so you understand where we come from.

In essence, our software is free, but our time isn't. I think we can all agree that it's in our best interest as users that we charge for our services and reinvest for a better product.

I understand that it's hard to justify getting a support contract for the occasional small issue that shouldn't take "more than 5 minutes to fix". But if we give out our software and support for free, what's left?

With this, I'd like to propose that this forum is here for the community to help itself, but that we avoid requests specifically made to the Scalr company or Scalr staff. We prefer a karma-based system where the more an individual answers others' questions, contributes to documentation, and adds value for others, and more likely we'll set time aside to investigate an issue, try to reproduce, or develop a fix for them.

If any of you know any community management software that would facilitate this, kind of like ohloh.net's kudos (http://meta.ohloh.net/kudos/), please let us know.

Thoughts?

[Srinivasan Subramanian]

Hi Sebastian

 

In principle there are no disconnects with what you are proposing below.  I am sure everyone appreciates the work your team is doing in providing this as Open Source and the need to put a business model around it.  Many successful companies have worked out this model and I am sure you will also be successful.

 

Having said that, here is my feedback.

 

1. I think that the model wherein when someone contributes from the community, your team will support those issues a little more is not appealing.  Its some how not democratic

2. What Scalr really needs is a rich, vibrant community.  Why do I say that?  Since the time I have been posting queries on the forum (last 3 or 4 months), very few questions have been answered by the community.  Only some questions have been answered and many have gone unanswered.  Either I have found some answers myself or they are still pending.  The few which were answered, were answered by the Scalr team.

3  The traffic is also not very high.

4. Take my last query wherein I appealed to your team for an answer.  I don’t expect a solution but only an answer so I can find some solutions myself.

5. The fact that there are very few answers from the community points not (in my view) to apathy for the product but more to either:

    a.  Fewer users using the product (or)

    b.  Limited knowledge amongst the community to answer.

6.  IMHO I think it’s a bit of both, but more of the second reason.  Scalr is not easy s/w to understand since its solving a complex problem.  The only way your team can focus more and more on product development and less on community support is by building up a vibrant community.  Many companies have a community manager and try to rally the community.

7.  The forums cannot survive only community contributions but also need internal Scalr help.  The help to be provided is answers more than fixes.  A good case in point is my query on whether there is an automated upgrade from 2.5 to 3.0, the answer was No and I found a way around it (and posted the info back to the forum to help others ).  What was important was that I got a quick answer and hence proceeded with the next step.  The challenge is that many times there are no answers.

 

How can this change?  I would request that you scout for volunteers to support your effort on the forums.  I volunteer readily for this effort.  Once you have some volunteers, please provide us some further info than what is already available so that questions on the forums can be answered by the volunteers.  This will gradually allow your team to focus lesser on the unpaid support and focus more on product advancement.

 

Its better to spend some effort and time in building up a community.  The answers and solutions will then automatically come.

 

Cheers

Srini

--
You received this message because you are subscribed to the Google Groups "scalr-discuss" group.
To view this discussion on the web visit https://groups.google.com/d/msg/scalr-discuss/-/XbQl4Ol9SnkJ.
To post to this group, send email to scalr-...@googlegroups.com.
To unsubscribe from this group, send email to scalr-discus...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/scalr-discuss?hl=en.

[mavinman]

+1 Srini. Well written, fully agree.

On Apr 4, 12:08 am, "Srinivasan Subramanian"

<srinivasan.subraman...@gmail.com> wrote:
> Hi Sebastian
>
> In principle there are no disconnects with what you are proposing below.  I am sure everyone appreciates the work your team is doing in providing this as Open Source and the need to put a business model around it.  Many successful companies have worked out this model and I am sure you will also be successful.
>
> Having said that, here is my feedback.
>
> 1. I think that the model wherein when someone contributes from the community, your team will support those issues a little more is not appealing.  Its some how not democratic

> 2. What Scalr really needs is a rich, vibrant community.  Why do I say that?  Since the time I have been posting queries on the forum (last 3 or 4 months), very few questions have been answered by the community.  Only some questions have been answered and many have gone unanswered.  Either I have found some answers myself or they are still pending.  The few which were answered, were answered by the Scalr team.
> 3  The traffic is also not very high.
> 4. Take my last query wherein I appealed to your team for an answer.  I don’t expect a solution but only an answer so I can find some solutions myself.
> 5. The fact that there are very few answers from the community points not (in my view) to apathy for the product but more to either:
>     a.  Fewer users using the product (or)
>     b.  Limited knowledge amongst the community to answer.
> 6.  IMHO I think it’s a bit of both, but more of the second reason.  Scalr is not easy s/w to understand since its solving a complex problem.  The only way your team can focus more and more on product development and less on community support is by building up a vibrant community.  Many companies have a community manager and try to rally the community.

> 7.  The forums cannot survive only community contributions but also need internal Scalr help.  The help to be provided is answers more than fixes.  A good case in point is my query on whether there is an automated upgrade from 2.5 to 3.0, the answer was No and I found a way around it (and posted the info back to the forum to help others ).  What was important was that I got a quick answer and hence proceeded with the next step.  The challenge is that many times there are no answers.

> To view this discussion on the web visithttps://groups.google.com/d/msg/scalr-discuss/-/XbQl4Ol9SnkJ.

> To post to this group, send email to scalr-...@googlegroups.com.
> To unsubscribe from this group, send email to scalr-discus...@googlegroups.com.
> For more options, visit this group athttp://groups.google.com/group/scalr-discuss?hl=en.
>

>  wlEmoticon-smile[1].png
> 1KViewDownload

[Juan Granados]

+2 Srini. You have helped me out alot. It would be nice to have more
community support. Scalr is a great application.

Juan

[Sebastian Stadil]

We have consensus on the problem: community users have questions, and some of those questions can only be answered by Scalr staff.

Where I believe we differ is the way to address this problem. You seem to suggest that we should indiscriminately answer them. We propose to selectively answer them based on some criteria (person's contribution to the community, ability to have the question answered by the community, time involved in answering).

I don't believe that indiscriminately answering questions will create a self-sustaining community forum. I believe encouraging people to post will do so better, especially if there is an incentive. Thoughts?

--

Follow us: Twitter - Facebook - Blog

[Srinivasan Subramanian]

Hi Sebastian

We are dealing with a chicken and egg problem   Scalr wants more community contribution.  While I don’t disagree, I am trying to highlight the problem that even if we want to help many times we are unable to help.

 

I leave it to your team to decide how to proceed.  All I have known and seen is that every great open source company always “seeds” community engagement.  The larger the community the more the contribution.  I don’t know the exact numbers, I am sure you know the spread of Scalr community.  Please put in place any model that you think will work, all I want (and hope) is that Scalr gets stronger.

> > To unsubscribe from this group, send email to mailto:scalr-discuss%2Bunsu...@googlegroups.com.

> > For more options, visit this group athttp://groups.google.com/group/scalr-discuss?hl=en.
>
> >  wlEmoticon-smile[1].png
> > 1KViewDownload

--
You received this message because you are subscribed to the Google Groups "scalr-discuss" group.
To post to this group, send email to scalr-...@googlegroups.com.

To unsubscribe from this group, send email to mailto:scalr-discuss%2Bunsu...@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/scalr-discuss?hl=en.

 

--

 

Follow us: Twitter - Facebook - Blog

 

[Sebastian Stadil]

We tried the seeding model, and all we got were freeloaders. If it's easy to get an answer, I don't see the incentive to help.

What are the use cases for deploying Scalr yourselves? Perhaps that will help me understand better where you are coming from.

[Srinivasan Subramanian]

The usual suspects I guess:

 

1. Cant afford the paid service

2. One month trial may be too less for someone to really get the hang of Scalr

3. Company policy needs everything running in house

4. All development / trial landscapes to be done in house, production using the paid service (maybe)

5.  Own data center and cannot expose the Eucalyptus or any other stack

6.  Its open source, so I wan't to use it and run it myself :)

7.  The standard service expects most people to use Scalr with LAMP stack or WAMP.  Similar to my use case, people could be using Scalr to run other stuff (ours is on JBoss) and may need lot more customizing.  The customizing could be just scripts or more involving changing of Scalr code, custom roles etc etc

 

That's the list I can think of to start with ...