Re: [Users] How to configure sharedFS ?

2012-06-17 Thread Deepak C Shetty

On 03/19/2012 03:07 PM, Dan Kenigsberg wrote:

On Wed, Mar 14, 2012 at 07:04:47PM +0530, Deepak C Shetty wrote:

On 02/29/2012 08:06 PM, Dan Kenigsberg wrote:

Let us know if you need further assistance..
If you have that python script working for LOCALFS, I'd suggest you try
making it work for SHAREDFS too.



I was able to use the VDSM_Standalone example, and modify it to use SHAREDFS
and export glusterfs as a DATA_DOMAIN and invoke VM backed by
gluster storage.

I edited http://www.ovirt.org/wiki/Vdsm_Standalone (scroll below)
added it as
a SHAREDFS example.

Thanks! It is great news that all you need is O_DIRECT in fuse, and that
Vdsm's SHAREDFS interface works as it is for gluster.

Regards,
Dan.



Hello,
Got more questions on this, now that I am re-visiting this.
The last time i tried using SHAREDFS, i started from createStorageDomain 
verb, and it works fine.


But now we have connectStorageServer verb... which i believe is the new 
way of doing things ?
If i start from connectStorageServer verb to mount using SHAREDFS ( 
which goes via PosixFs... MountConnection flow), that won't help me 
entirely here, right ?
Because it only mounts based on the dict sent, but does not do anything 
with the image and metadata stuff ( which createStorageDomain flow did ).


I am wondering if its too early to start using connectStorageServer ? If 
not, how can i re-write the above vdsm standalone example using 
connectStorageServer instead of createStorageDomain flow ?



___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users


Re: [Users] Following Up on LinuxCon Japan Workshop

2012-06-17 Thread Barak Azulay

On 06/17/2012 10:00 AM, Yair Zaslavsky wrote:

On 06/14/2012 06:02 PM, Barak Azulay wrote:

On 06/14/2012 02:41 AM, Mike Burns wrote:

Hi Leslie,

On Wed, 2012-06-13 at 15:14 -0700, Leslie Hawthorn wrote:

Hello everyone,

I'd like to collect feedback from everyone on last week's oVirt workshop
at LinuxCon Japan. [0] Please reply back with comments by close of
business this Friday, 15 June.

Once we have gathered feedback on list, I'll capture it on the oVirt
wiki. We can then use what we've learned to help construct the agenda
and other plans for other upcoming oVirt workshops at LinuxCons. [1]

If you know that you would like to volunteer as an instructor for future
workshops or would like to suggest alternate content for the workshop,
please include that in your feedback.

- Course Material
What sessions were most well received? Which ones require improvement?
Any additional sessions we'd suggest?


All of the sessions seemed to be well received.  I don't know that any
need blatant improvement, though they could probably all take a bit of
slide updates.  The ovirt-node presentation was especially out of date
(my fault) with it's screenshots.




I would like to add:
- a demo is a must in a workshop, as it is the most effective and
   impressive way to attract people.
- I think the ovirt-engine presentation needs a bit more meat

What stuff do you think we need to add to this presentation?
Can you be more specific?
Thanks,

Yair



- better diagrams
- code structure
- go over a simple example-command work  flow - from entry till DB ...

and many more






- I would also add vertical presentations:
 * ovirt network
 * ovirt storage
 * ovirt vm life cycle
   Of course all the above should be presented after the arch  demo
   presentation







- Audience Participation
How many attendees? How did the QA periods go? Would we like to prepare
a post-event attendee survey? (I recommend we do the survey and can
prepare some questions for the group if that's useful.)


There were probably between 25-35 people through the course of the day
for all the presentations.



I gave also an intro/arch talk also in the virt mini-summit, I actually
gave 2.
I volunteered to cover for one of the presenters that surprisingly
didn't show up on time, because of the confusion this session was rather
small (~40)
In the second one (in my scheduled slot), there were about ~80.





There wasn't really a lot of time set aside for QA aside from a few
minutes at the end of each presentation and the breaks in between
presentations.  During those times, there generally were people asking
questions.



Actually I wasn't expecting too many questions due to the different
culture of the audience, but we did get questions and people showed
interest.
Most of the interesting questions came during the breaks, and I had a
few white-board sessions with those people.





I think we should prepare the survey in advance of the next workshop and
ask people to fill it out during the workshop, rather than coming back a
couple weeks after the fact.  We didn't really collect names of the
people in attendance and I expect that a fair number of people
registered attended the Gluster Conference.



The linux foundation scanned all the participants tags, so I guess they
can supply the names.




- Developer/user traction resulting from engaging at workshop
Did this workshop help us to gain new developers or users? Reinforce
relationships with existing community members?


Only time will really tell here.  People definitely seemed interested
and there were a couple people already downloading and installing the
bits during the workshop.  I got the impression that most of the people
there were more corporate focused -- Hitachi and Fujitsu were both well
represented.  They're from companies that have previously expressed
interest in the project.




Hard to tell,
I had a feeling that there was also a none developers group that came
also to learn about ovirt, more of the consultant type that actually
considering ovirt as an option.




- Promotion of the event, both before and after
What could be done to more effectively promote the event prior to the
workshops? Videos and slides from the workshop should be posted on the
LC Japan site tomorrow; what action would the community like to take to
promote this content?


Not really my area of expertise -- promoting.  Probably better to get
information out there earlier that the workshop is happening.  It seemed
to be a bit of a fire drill getting things together for this workshop
since it was only about a month before that we decided to run it.
Future workshops shouldn't be as big a deal.




I would say we need a lot of promotion, I was actually accessed by a few
people after the into presentation, saying they had no idea that ovirt
is so mature and feature reach.
This message should be conveyed before the workshop.




Probably an email announcement to users@ with a link to the
presentations/videos 

Re: [Users] How to configure sharedFS ?

2012-06-17 Thread Deepak C Shetty

On 06/17/2012 07:06 PM, Yair Zaslavsky wrote:

On 06/17/2012 03:50 PM, Deepak C Shetty wrote:

On 03/19/2012 03:07 PM, Dan Kenigsberg wrote:

On Wed, Mar 14, 2012 at 07:04:47PM +0530, Deepak C Shetty wrote:

On 02/29/2012 08:06 PM, Dan Kenigsberg wrote:

Let us know if you need further assistance..
If you have that python script working for LOCALFS, I'd suggest you try
making it work for SHAREDFS too.



I was able to use the VDSM_Standalone example, and modify it to use
SHAREDFS
and export glusterfs as a DATA_DOMAIN and invoke VM backed by
gluster storage.

I edited http://www.ovirt.org/wiki/Vdsm_Standalone (scroll below)
added it as
a SHAREDFS example.

Thanks! It is great news that all you need is O_DIRECT in fuse, and that
Vdsm's SHAREDFS interface works as it is for gluster.

Regards,
Dan.


Hello,
 Got more questions on this, now that I am re-visiting this.
The last time i tried using SHAREDFS, i started from createStorageDomain
verb, and it works fine.

But now we have connectStorageServer verb... which i believe is the new
way of doing things ?
If i start from connectStorageServer verb to mount using SHAREDFS (
which goes via PosixFs... MountConnection flow), that won't help me
entirely here, right ?
Because it only mounts based on the dict sent, but does not do anything
with the image and metadata stuff ( which createStorageDomain flow did ).

I am wondering if its too early to start using connectStorageServer ? If
not, how can i re-write the above vdsm standalone example using
connectStorageServer instead of createStorageDomain flow ?

Will a solution via engine (i.e - create PosixFS storage domain) be good
for you?


Well, no. I kind of mis-interpreted the problem. My Q stands canceled.
The right way indeed is to use createStorageServer followed by 
createStorageDomain, which i was confused initially a bit, but now its 
cleared.



___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users