bug in gocept.zeoraid:server recipe

[ChrisW]

I think the same problem affects zc.zodbrecipes:server, but since the problem I bumped into with was with the zeoraid recipe, I'll report here...

So, I have the following section:

[zeoraid1]
recipe = gocept.zeoraid:server
zeo.conf =
    <zeo>
    address 127.0.0.1:8200
    </zeo>
    %import gocept.zeoraid
    <raidstorage main>
        <zeoclient 1>
            server localhost:8100
            storage 1
        </zeoclient>
    </raidstorage>

Steps to reproduce:

1. start the zeoraid server

2. edit the above config

3. run bin/buildout

4. attempt to stop the zeoraid server

Because step 3 blows away the part directory, the .pid file is lost and step 4 fails because the control script no longer thinks zeoraid is running.

How do I get this recipe to put its .pid files somewhere less volatile?
(The bug is that neither gocept.zeoraid:server nor zc.zodbrecipes:server do this by default)

[Christian Theune (ctheune)]

On 10/30/2009 12:50 AM, ChrisW wrote:
> Public bug reported:
>
> I think the same problem affects zc.zodbrecipes:server, but since the
> problem I bumped into with was with the zeoraid recipe, I'll report
> here...
>
> So, I have the following section:
>
> [zeoraid1]
> recipe = gocept.zeoraid:server
> zeo.conf =
> <zeo>
> address 127.0.0.1:8200
> </zeo>
> %import gocept.zeoraid
> <raidstorage main>
> <zeoclient 1>
> server localhost:8100
> storage 1
> </zeoclient>
> </raidstorage>
>
> Steps to reproduce:
>
> 1. start the zeoraid server
>
> 2. edit the above config
>
> 3. run bin/buildout
>
> 4. attempt to stop the zeoraid server
>
> Because step 3 blows away the part directory, the .pid file is lost and
> step 4 fails because the control script no longer thinks zeoraid is
> running.
>
> How do I get this recipe to put its .pid files somewhere less volatile?
> (The bug is that neither gocept.zeoraid:server nor zc.zodbrecipes:server do this by default)

That can be achieved using deployment recipes, the two I've used before are:

zc.recipe.deployment
gocept.recipe.deploymentsandbox

Does that work for you?

Christian

Status: Feedback required
Importance: Normal

--
Christian Theune · <email address hidden>
gocept gmbh & co. kg · forsterstraße 29 · 06112 halle (saale) · germany
http://gocept.com · tel [phone] · fax [phone]
Zope and Plone consulting and development

[Christian Theune (ctheune)]

Not an issue of the recipe itself. Use a complementing deployment recipe.

[ChrisW (chris-simplistix)]

Hmm... at the very least it's a documentation issue, a big warning is needed to stop people inadvertently losing their data.
Myself, I think the deployment option should be mandatory if the recipe is going to shoot you in the foot if you don't use one...

[Christian Theune (ctheune)]

Fixed in revision 105638. I hope this comment is sufficient.

[ChrisW (chris-simplistix)]

Given the possibility of destroying data and causing confusion, I still think the "deployment" should be mandatory.
However, I know Jim was having a think about the same issue in zc.zodbrecipes. Might be worth having a chat with him.

I've re-opened this issue, but I think the Importance on this issue could be set as Low...