Mailing-List: contact dev-help@cloudstack.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cloudstack.apache.org
Content-Type: multipart/alternative;
 boundary="===============5455427067266380240=="
MIME-Version: 1.0
Subject: Re: Review Request: Delete KVMHA dir contents if found when deleting
 templates, volumes from nfs secondary storage
From: "Marcus Sorensen" <shadowsor@gmail.com>
To: "Chip Childers" <chipchilders@apache.org>
Cc: "Marcus Sorensen" <shadowsor@gmail.com>,
 "cloudstack" <cloudstack-dev@incubator.apache.org>,
 "edison su" <edison.su@citrix.com>
Date: Thu, 25 Apr 2013 00:11:23 -0000
Message-ID: <20130425001123.22548.45764@reviews.apache.org>
Auto-Submitted: auto-generated
Sender: "Marcus Sorensen" <noreply@reviews.apache.org>
References: <20130424230542.22548.16302@reviews.apache.org>
In-Reply-To: <20130424230542.22548.16302@reviews.apache.org>
Reply-To: "Marcus Sorensen" <shadowsor@gmail.com>

--===============5455427067266380240==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable


> On April 24, 2013, 11:05 p.m., edison su wrote:
> > The HA will write data into secondary storage also? It should only writ=
e data into primary storage.
> =

> Marcus Sorensen wrote:
>     The KVMStoragePoolManager is currently adding any NFS type to the lis=
t of HA pools to write to. I can change that as well, I'm just not very fam=
iliar with why it was working that way in the first place. Will look at it =
now.

Ok, I think I know what the issue is. In 4.0, primary storage pools were cr=
eated via KVMStoragePoolManager's createStoragePool, which calls LibvirtSto=
rageAdaptor's createStoragePool, and then adds it to the HA list. Secondary=
 storage shortcuts directly into LibvirtStorageAdaptor's createStoragePool,=
 and never gets added to the HA list. In 4.1, everything goes through KVMSt=
oragePool so we can identify the proper storage adaptor. I see no simple wa=
y to differentiate primary vs secondary NFS storage from within KVMStorageP=
oolManager, other than the 'template/tmpl', 'volumes', and 'snapshots' dire=
ctories in their paths. We may have to pass that in.


- Marcus


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/10766/#review19676
-----------------------------------------------------------


On April 24, 2013, 9:23 p.m., Marcus Sorensen wrote:
> =

> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/10766/
> -----------------------------------------------------------
> =

> (Updated April 24, 2013, 9:23 p.m.)
> =

> =

> Review request for cloudstack and Chip Childers.
> =

> =

> Description
> -------
> =

> the KVM HA runner uses any NFS secondary storage resource available to a =
host to store it's HA data. This causes template deletes to fail because it=
 cannot delete KVMHA, which is a directory that is not empty. So if KVMHA d=
irectory is found, delete it's contents before trying to delete it.
> =

> =

> This addresses bug CLOUDSTACK-2173.
> =

> =

> Diffs
> -----
> =

>   core/src/com/cloud/storage/resource/NfsSecondaryStorageResource.java e6=
5cbe1 =

> =

> Diff: https://reviews.apache.org/r/10766/diff/
> =

> =

> Testing
> -------
> =

> Had a bunch of templates that the scavenger was failing to clean up. Appl=
ied this patch, installed the code, and the templates all cleaned up succes=
sfully.
> =

> =

> Thanks,
> =

> Marcus Sorensen
> =

>


--===============5455427067266380240==--