Gluster volume & system vms not starting

[Brent Hartzell <br...@outlook.com>]

Hello,

 

We are currently testing out CloudStack and continue to run into issues.

 

We have a physical server with CS management running on it & two hosts using
1 drive as an OS and the remaining 3 drives per host as a Gluster volume.

 

If we run the install of CloudStack per the guide, and use basic NFS for
storage, everything works. As soon as we try to use Gluster, CS will see the
storage, report it correctly, but nothing works - system vm's go into a loop
of starting-stopped and no vm's can be provisioned. 

 

The log keeps showing "Primary secondary storage is not even started" - not
really anything else that makes any sense in the logs though..

 

We've tried this with using both a shared mount point on the hosts as well
as using the Gluster option in primary storage setup. Both report the
correct amount of storage, but it's basically unusable since these system
VM's can't start.

 

That said, it seems like these system vm's are a single point of failure -
we had a handful of VM's running for testing on NFS on the management server
- if anything gets jacked up on those system vm's, it appears to essentially
take all of the VM's offline. Unless we had something configured wrong. Is
there any documentation on correctly using Gluster with CS?

 

 

 

 

RE: Gluster volume & system vms not starting

[Vadim Kimlaychuk <Va...@Elion.ee>]

Hi,

Check if you hypervisor host can access gluster storage. Try to mount it at hypervisor manually and check the files. We have used CS with Ceph and seing no problem to set it up as primary storage (in addition to NFS). 

Vadim.

-----Original Message-----
From: Brent Hartzell [mailto:brent.hartzell@outlook.com] 
Sent: Monday, October 27, 2014 12:13 AM
To: users@cloudstack.apache.org
Subject: Gluster volume & system vms not starting

Hello,

 

We are currently testing out CloudStack and continue to run into issues.

 

We have a physical server with CS management running on it & two hosts using
1 drive as an OS and the remaining 3 drives per host as a Gluster volume.

 

If we run the install of CloudStack per the guide, and use basic NFS for storage, everything works. As soon as we try to use Gluster, CS will see the storage, report it correctly, but nothing works - system vm's go into a loop of starting-stopped and no vm's can be provisioned. 

 

The log keeps showing "Primary secondary storage is not even started" - not really anything else that makes any sense in the logs though..

 

We've tried this with using both a shared mount point on the hosts as well as using the Gluster option in primary storage setup. Both report the correct amount of storage, but it's basically unusable since these system VM's can't start.

 

That said, it seems like these system vm's are a single point of failure - we had a handful of VM's running for testing on NFS on the management server
- if anything gets jacked up on those system vm's, it appears to essentially take all of the VM's offline. Unless we had something configured wrong. Is there any documentation on correctly using Gluster with CS?

 

 

 

 

Re: Gluster volume & system vms not starting

[Nux! <nu...@li.nux.ro>]

Please send us logs from the mgmt and HVs and make sure to enable DEBUG mode in the loggers, e.g. "sed -i 's/INFO/DEBUG/g' /etc/cloudstack/agent/log4j-cloud.xml" and restart..

Lucian

--
Sent from the Delta quadrant using Borg technology!

Nux!
www.nux.ro

----- Original Message -----
> From: "Brent Hartzell" <br...@outlook.com>
> To: users@cloudstack.apache.org
> Sent: Sunday, 26 October, 2014 22:12:47
> Subject: Gluster volume & system vms not starting

> Hello,
> 
> 
> 
> We are currently testing out CloudStack and continue to run into issues.
> 
> 
> 
> We have a physical server with CS management running on it & two hosts using
> 1 drive as an OS and the remaining 3 drives per host as a Gluster volume.
> 
> 
> 
> If we run the install of CloudStack per the guide, and use basic NFS for
> storage, everything works. As soon as we try to use Gluster, CS will see the
> storage, report it correctly, but nothing works - system vm's go into a loop
> of starting-stopped and no vm's can be provisioned.
> 
> 
> 
> The log keeps showing "Primary secondary storage is not even started" - not
> really anything else that makes any sense in the logs though..
> 
> 
> 
> We've tried this with using both a shared mount point on the hosts as well
> as using the Gluster option in primary storage setup. Both report the
> correct amount of storage, but it's basically unusable since these system
> VM's can't start.
> 
> 
> 
> That said, it seems like these system vm's are a single point of failure -
> we had a handful of VM's running for testing on NFS on the management server
> - if anything gets jacked up on those system vm's, it appears to essentially
> take all of the VM's offline. Unless we had something configured wrong. Is
> there any documentation on correctly using Gluster with CS?