To explain a bit further, currently system metrics collects some of its data from the opendaylight TSDR project, which is (AFAIK) no longer maintained. We have spoken about writing a new metrics collector to replace TSDR, for use by the system metrics backend. I was planning on doing this work sometime early next year once my schedule clears up a bit, but I’m certainly open to turning it into an intern project instead, and I would even be willing to oversee/provide guidance/etc for whomever is willing to expend some cycles on this.

 

/t

 

From: <systemmetrics-dev-bounces@...> on behalf of "Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco)" <dmalacho@...>
Date: Wednesday, December 14, 2016 at 10:45
To: "Lorand Jakab -X (lojakab - AAP3 INC at Cisco)" <lojakab@...>, Luis Gomez <ecelgp@...>, Alexis de Talhouët <adetalhouet@...>
Cc: controller-dev <controller-dev@...>, "integration-dev@..." <integration-dev@...>, "systemmetrics-dev@..." <systemmetrics-dev@...>
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

I don't understand this part. You finding a replacement would we the interesting part?

 

I meant, that removing TSDR dependency from System Metrics app and develop something different may be interesting topic for an intern.

 

dano

 

From: Lorand Jakab -X (lojakab - AAP3 INC at Cisco)
Sent: 13. decembra 2016 22:07
To: Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco); Luis Gomez; Alexis de Talhouët
Cc: integration-dev@...; controller-dev; systemmetrics-dev@...
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

Hi Dano,

Thanks for the info. See inline...

On 12/13/16 10:10 PM, Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco) wrote:

Hi,

 

Code for the project will be in the repo in few days, we had some delay in repository creation J

 

About your proposal Lori.

Lot of what you wrote is done now – reading cluster members, shards, some shard metrics etc.

Yes, after I saw the screenshots from the project proposal, that was pretty much what I was looking for.

 

This looks interesting:

“In addition to exposing cluster information, the app would also allow activating a failover from an active cluster in one location to a backup cluster in another location, by manipulating voting states of shards.”

This was suggested by Tom Pantelis at the last ODL summit, the idea is to have a 3-node cluster in one location as the active cluster, and another 3-node cluster in a different datacenter for backup. Since inter-datacenter latency can be high, the backup nodes should be non-voting, so they can't become leaders and they don't hold up transactions. Maybe this could be integrated somewhere in the UI.

 

Also, we are using TSDR for “harvesting” controller and machine metrics from JVM Console (I think) in the backend part and we are planning to remove it.

Maybe this can be interesting too…

I don't understand this part. You finding a replacement would we the interesting part?

-Lori

 

 

dano

 

From: systemmetrics-dev-bounces@... [mailto:systemmetrics-dev-bounces@...] On Behalf Of Luis Gomez
Sent: 13. decembra 2016 19:32
To: Alexis de Talhouët
Cc: integration-dev@...; controller-dev; Lorand Jakab -X (lojakab - AAP3 INC at Cisco); systemmetrics-dev@...
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

+1, this project if still alive could leverage the intern work.

 

On Dec 13, 2016, at 5:56 AM, Alexis de Talhouët <adetalhouet@...> wrote:

 

 

On Dec 13, 2016, at 8:55 AM, Lori Jakab <lojakab@...> wrote:

 

On 12/13/16 3:53 PM, Lori Jakab wrote:

Thanks Alexis, this is useful information. There seems to be some
overlap with my project proposal, I will take the repo for a spin, see
what's supported for now.

Actually, while the repository was created, initial code was not
submitted yet.

 

Correct, still no code there: https://git.opendaylight.org/gerrit/#/q/project:systemmetrics

 

I beleive you could sync up with the system metrics team to better assess their scope, and what the internship proposal could add to that.

-Lori

-Lori

On 12/13/16 3:51 PM, Alexis de Talhouët wrote:

Acutally, I might have introduced a confusion. The repo and the
project are called System Metrics, not Cluster Metrics as proposed in
a first place.

This is the current
repository: https://github.com/opendaylight/systemmetrics

That project was accepted the 2016-11-10 as an incubation project by
TSC. + systemmetrics-dev mailing list (although I'm not sure a lot of
people are already on it).

Hope this helps,
Alexis

On Dec 13, 2016, at 8:07 AM, Alexis de Talhouët
<adetalhouet@... <mailto:adetalhouet@...>> wrote:

Hey Lori,

I can't step up as a co-mentordue to lack of knowledge about
clustering, but regarding the portal aggregating data, a new project
called Cluster Metrics [0] was recently accepted. I believe the scope
of the proposed internship could be coupled with the functionalities
that project is going to provide to build a more cohersive
"Dashboard" of the cluster along with some cool functionalities for
the shards.

Regards,
Alexis

[0]: https://wiki.opendaylight.org/view/Project_Proposals:Cluster_Metrics

On Dec 13, 2016, at 7:59 AM, Lori Jakab <lojakab@...
<mailto:lojakab@...>> wrote:

Hi,

I'm far from being an expert on ODL clustering, but I proposed a
clustering related internship project [0], because I recently explored a
bit the topic, and came up with some ideas to improve the status quo. We
will execute on them, but get the best possible results, I would really
like to get a volunteer who is more involved with and more knowledgeable
about clustering than I am, to co-mentor this project with me. You can
read more about the project at the link [0], but the gist of it is to
expose as much information about a cluster as possible in a DLUX app,
and allow active/backup fail-over from one cluster to another one, at
the geographically remote location (which implies that shards on the
backup cluster are initially in non-voting mode, and then a switch is
flipped for the fail-over).

Any takers?

Thanks,
-Lori

[0]
https://wiki.opendaylight.org/view/Interns/Projects#Portal_for_Cluster_Management
_______________________________________________
integration-dev mailing list
integration-dev@...
<mailto:integration-dev@...>
https://lists.opendaylight.org/mailman/listinfo/integration-dev

 

 

_______________________________________________
integration-dev mailing list
integration-dev@...
https://lists.opendaylight.org/mailman/listinfo/integration-dev

 

 

To explain a bit further, currently system metrics collects some of its data from the opendaylight TSDR project, which is (AFAIK) no longer maintained. We have spoken about writing a new metrics collector to replace TSDR, for use by the system metrics backend. I was planning on doing this work sometime early next year once my schedule clears up a bit, but I’m certainly open to turning it into an intern project instead, and I would even be willing to oversee/provide guidance/etc for whomever is willing to expend some cycles on this.

 

/t

 

From: <systemmetrics-dev-bounces@...> on behalf of "Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco)" <dmalacho@...>
Date: Wednesday, December 14, 2016 at 10:45
To: "Lorand Jakab -X (lojakab - AAP3 INC at Cisco)" <lojakab@...>, Luis Gomez <ecelgp@...>, Alexis de Talhouët <adetalhouet@...>
Cc: controller-dev <controller-dev@...>, "integration-dev@..." <integration-dev@...>, "systemmetrics-dev@..." <systemmetrics-dev@...>
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

I don't understand this part. You finding a replacement would we the interesting part?

 

I meant, that removing TSDR dependency from System Metrics app and develop something different may be interesting topic for an intern.

 

dano

 

From: Lorand Jakab -X (lojakab - AAP3 INC at Cisco)
Sent: 13. decembra 2016 22:07
To: Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco); Luis Gomez; Alexis de Talhouët
Cc: integration-dev@...; controller-dev; systemmetrics-dev@...
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

Hi Dano,

Thanks for the info. See inline...

On 12/13/16 10:10 PM, Daniel Malachovsky -X (dmalacho - PANTHEON TECHNOLOGIES at Cisco) wrote:

Hi,

 

Code for the project will be in the repo in few days, we had some delay in repository creation J

 

About your proposal Lori.

Lot of what you wrote is done now – reading cluster members, shards, some shard metrics etc.

Yes, after I saw the screenshots from the project proposal, that was pretty much what I was looking for.

 

This looks interesting:

“In addition to exposing cluster information, the app would also allow activating a failover from an active cluster in one location to a backup cluster in another location, by manipulating voting states of shards.”

This was suggested by Tom Pantelis at the last ODL summit, the idea is to have a 3-node cluster in one location as the active cluster, and another 3-node cluster in a different datacenter for backup. Since inter-datacenter latency can be high, the backup nodes should be non-voting, so they can't become leaders and they don't hold up transactions. Maybe this could be integrated somewhere in the UI.

 

Also, we are using TSDR for “harvesting” controller and machine metrics from JVM Console (I think) in the backend part and we are planning to remove it.

Maybe this can be interesting too…

I don't understand this part. You finding a replacement would we the interesting part?

-Lori

 

 

dano

 

From: systemmetrics-dev-bounces@... [mailto:systemmetrics-dev-bounces@...] On Behalf Of Luis Gomez
Sent: 13. decembra 2016 19:32
To: Alexis de Talhouët
Cc: integration-dev@...; controller-dev; Lorand Jakab -X (lojakab - AAP3 INC at Cisco); systemmetrics-dev@...
Subject: Re: [Systemmetrics-dev] [integration-dev] Looking for a co-mentor for internship project

 

+1, this project if still alive could leverage the intern work.

 

On Dec 13, 2016, at 5:56 AM, Alexis de Talhouët <adetalhouet@...> wrote:

 

 

On Dec 13, 2016, at 8:55 AM, Lori Jakab <lojakab@...> wrote:

 

On 12/13/16 3:53 PM, Lori Jakab wrote:

Thanks Alexis, this is useful information. There seems to be some
overlap with my project proposal, I will take the repo for a spin, see
what's supported for now.

Actually, while the repository was created, initial code was not
submitted yet.

 

Correct, still no code there: https://git.opendaylight.org/gerrit/#/q/project:systemmetrics

 

I beleive you could sync up with the system metrics team to better assess their scope, and what the internship proposal could add to that.

-Lori

-Lori

On 12/13/16 3:51 PM, Alexis de Talhouët wrote:

Acutally, I might have introduced a confusion. The repo and the
project are called System Metrics, not Cluster Metrics as proposed in
a first place.

This is the current
repository: https://github.com/opendaylight/systemmetrics

That project was accepted the 2016-11-10 as an incubation project by
TSC. + systemmetrics-dev mailing list (although I'm not sure a lot of
people are already on it).

Hope this helps,
Alexis

On Dec 13, 2016, at 8:07 AM, Alexis de Talhouët
<adetalhouet@... <mailto:adetalhouet@...>> wrote:

Hey Lori,

I can't step up as a co-mentordue to lack of knowledge about
clustering, but regarding the portal aggregating data, a new project
called Cluster Metrics [0] was recently accepted. I believe the scope
of the proposed internship could be coupled with the functionalities
that project is going to provide to build a more cohersive
"Dashboard" of the cluster along with some cool functionalities for
the shards.

Regards,
Alexis

[0]: https://wiki.opendaylight.org/view/Project_Proposals:Cluster_Metrics

On Dec 13, 2016, at 7:59 AM, Lori Jakab <lojakab@...
<mailto:lojakab@...>> wrote:

Hi,

I'm far from being an expert on ODL clustering, but I proposed a
clustering related internship project [0], because I recently explored a
bit the topic, and came up with some ideas to improve the status quo. We
will execute on them, but get the best possible results, I would really
like to get a volunteer who is more involved with and more knowledgeable
about clustering than I am, to co-mentor this project with me. You can
read more about the project at the link [0], but the gist of it is to
expose as much information about a cluster as possible in a DLUX app,
and allow active/backup fail-over from one cluster to another one, at
the geographically remote location (which implies that shards on the
backup cluster are initially in non-voting mode, and then a switch is
flipped for the fail-over).

Any takers?

Thanks,
-Lori

[0]
https://wiki.opendaylight.org/view/Interns/Projects#Portal_for_Cluster_Management
_______________________________________________
integration-dev mailing list
integration-dev@...
<mailto:integration-dev@...>
https://lists.opendaylight.org/mailman/listinfo/integration-dev

 

 

_______________________________________________
integration-dev mailing list
integration-dev@...
https://lists.opendaylight.org/mailman/listinfo/integration-dev