Stackato feedback

[Andy McKay]

Next week I’m meeting with Activestate peeps over a fine cup of tea to discuss Stackato. They are especially interested in how Stackato works from a *developer perspective*, not an operational perspective. I’ve got one or two emails from people on this subject, but feel free to tell me more. I think Stackato is pretty cool and mostly like using it, but then I get those frustrating times when it all goes wrong…

So, is it great? Does it suck? Tell me why and I’ll pass them along.

Andy

[Ralph Giles]

On 2014-06-06 10:01 AM, Andy McKay wrote:

> Next week I�m meeting with Activestate peeps over a fine cup of tea to discuss Stackato. They are especially interested in how Stackato works from a *developer perspective*, not an operational perspective. I�ve got one or two emails from people on this subject, but feel free to tell me more. I think Stackato is pretty cool and mostly like using it, but then I get those frustrating times when it all goes wrong�

I've only used our paas instance, so I can't really separate the two.

The first thing which struck me is the openness of the developer tool.
There's a 'stackato-cli' project on github, but it appears to need a
proprietary toolchain to build it? I couldn't figure out if it was
practically open source or not.

This is a problem: it's a speed bump to getting started and raises trust
and maintenance issues. Binaries are one thing, but I should be able to
'apt-get install' or 'brew install' stackato and get a working client
environment.

Otherwise I hit several bugs in my small experiment. Websockets don't
work, and it seemed easy to confuse the runner if there were missing
dependencies or other errors in the manifest, such that push, update,
start/stop/restart would fail in unexpected ways. The service shouldn't
be an additional source of issues when debugging a simple project.

Things may be better now; our instance is running 2.10 instead of
current stable 3.2

That's my first-run impressions, fwiw.

-r

[Chris Turra]

On 6/9/14, 1:26 PM, Ralph Giles wrote:
> Otherwise I hit several bugs in my small experiment. Websockets don't
> work, and it seemed easy to confuse the runner if there were missing
> dependencies or other errors in the manifest, such that push, update,
> start/stop/restart would fail in unexpected ways. The service shouldn't
> be an additional source of issues when debugging a simple project.

the websockets issues is not a stackato one, it was a limitation with
the version of the load balancer [1] we were running. this is something
we should be able to sort out now after taking the last tree closing
window as an opportunity to upgrade all the load balancers in our infra
(that was a LONG day). surely there will be reconfiguration/testing
required, but my point is it should be possible now.


-cturra

[1] stringray by riverbed

[Kumar McMillan]

On Jun 6, 2014, at 12:01 PM, Andy McKay <amc...@mozilla.com> wrote:

> Next week I’m meeting with Activestate peeps over a fine cup of tea to discuss Stackato. They are especially interested in how Stackato works from a *developer perspective*, not an operational perspective. I’ve got one or two emails from people on this subject, but feel free to tell me more. I think Stackato is pretty cool and mostly like using it, but then I get those frustrating times when it all goes wrong…
>
> So, is it great? Does it suck? Tell me why and I’ll pass them along.

Not sure if the feedback meeting is still pending or not but one thing I’d like to do is just say ‘stackato push’ and have it do all the things without asking me any questions. There is probably already a way to do this -- maybe with some config file settings? Having to read the questions and type enter is mildly annoying and seems unnecessary.

>
> Andy
>
>
>
> _______________________________________________
> paas-users mailing list
> paas-...@mozilla.org
> https://mail.mozilla.org/listinfo/paas-users

[Christopher Van]

stackato push --no-prompt

_______________________________________________
dev-webdev mailing list
dev-w...@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-webdev

[Stefan Arentz]

On Jun 6, 2014, at 1:01 PM, Andy McKay <amc...@mozilla.com> wrote:

> Next week I’m meeting with Activestate peeps over a fine cup of tea to discuss Stackato. They are especially interested in how Stackato works from a *developer perspective*, not an operational perspective. I’ve got one or two emails from people on this subject, but feel free to tell me more. I think Stackato is pretty cool and mostly like using it, but then I get those frustrating times when it all goes wrong…
>
> So, is it great? Does it suck? Tell me why and I’ll pass them along.

It would be great to see support for the Go runtime. I see it mentioned in the official Stackato docs but I don’t think we have it?

And newer versions of MongoDB, Redis and RabbitMQ. The versions we have are not bad but they are a couple of versions behind the production releases.

Something that I really miss is corncobs. (Ha, autocorrect did that and I am just going to leave it :-) I often have apps that need to periodically run a job and AFAIK there is no cron-like functionality in Stackato. Or at least not in our instance. Specifically asking for Python apps, which do not have a persistent runtime. (Compared to for example a Go or Node one, where you could hack something in the main runloop to schedule things)

S.

[Andy McKay]

On Jun 12, 2014, at 1:21 PM, Stefan Arentz <sar...@mozilla.com> wrote:

>
> On Jun 6, 2014, at 1:01 PM, Andy McKay <amc...@mozilla.com> wrote:
>
>> Next week I’m meeting with Activestate peeps over a fine cup of tea to discuss Stackato. They are especially interested in how Stackato works from a *developer perspective*, not an operational perspective. I’ve got one or two emails from people on this subject, but feel free to tell me more. I think Stackato is pretty cool and mostly like using it, but then I get those frustrating times when it all goes wrong…
>>
>> So, is it great? Does it suck? Tell me why and I’ll pass them along.
>
> It would be great to see support for the Go runtime. I see it mentioned in the official Stackato docs but I don’t think we have it?

We don’t. To get a list of what we do support, run the command: stackato runtimes

> And newer versions of MongoDB, Redis and RabbitMQ. The versions we have are not bad but they are a couple of versions behind the production releases.

I don’t think those things are necessarily problems with the platform from Stackato’s point of view.

> Something that I really miss is corncobs. (Ha, autocorrect did that and I am just going to leave it :-) I often have apps that need to periodically run a job and AFAIK there is no cron-like functionality in Stackato. Or at least not in our instance. Specifically asking for Python apps, which do not have a persistent runtime. (Compared to for example a Go or Node one, where you could hack something in the main runloop to schedule things)

Crontabs are documented here: http://api.paas.allizom.org/docs/deploy/index.html?highlight=cron#crontab-support

Thanks for the feedback everyone, I’m meeting with them next week. I’ll let you know what happens.

Andy