Archive for the 'Dev Jam' Category

2014 Dev Jam – Day 6

Saturday, June 7th, 2014

Friday was pretty much the “Dev Jam Results Show”. Mike suggested that the various teams get together and present their work. He recorded it live from his iPhone using Ustream and the raw (and I mean raw) video is available for your viewing pleasure.

The first stream covered the following topics.

Ben, Ron and Matt R. – AngularJS based webUI

That OpenNMS could use a new user interface is a given, and the decision was made to base it on a technology called AngularJS. The first demo shows off some of the work that was done to build a framework for the new GUI.

This includes a “plugin” architecture that will make it easy to add functionality to the system as well as to embed existing code that, while not written in Angular, is still useful. Ron even managed to get KSC reports to run under the new code, complete with dynamic updates.

Work that still remains to be done include formal authentication (currently the new GUI just gets a session cookie from the old one) as well as a greater granularity for ReST permissions, as now normal users get a lot of data and admin users get all the data. This would be very useful for things like multi-tenancy.

Alejandro – Requistion Manager

When we wrote the provisioner, we knew we had something special as no other management system seems to take discovery as seriously as OpenNMS. As more and more people find novel ways of using this system, we realized that the user interface could use some improvement. In this section Alejandro demonstrates the changes he has made to the interface for creating and managing requisitions, also built on Angular.

Craig Gallen – High Frequency Trader GUI

Those of you that follow OpenNMS in the news might have seen a press release a couple of weeks ago from a financial trading services company called TMX Atrium Networks. Dr. Gallen, our man in the UK, worked with them to build an interface for monitoring latency across the network.

Matt and Eric – newts

Yesterday I talked about the New Time Series database that we are building as the data storage backend for OpenNMS. I lifted a lot of that from this portion of the demonstrations. The ability to have an incredibly fast and highly scalable data store is key for our goal of making OpenNMS the de facto network management platform of choice.

The second video stream features a talk about “snee-po”

Seth – SNMnepO

SNMnepO, or OpenNMS spelled backwards, is a project to create a distributed data collector with horizontal scalability. The idea is to add data collection to our remote poller, and Seth’s demonstration shows data collection being performed with the collectd process disabled.

Coupled with the newts data storage backend, this new distributed collector will insure that OpenNMS can scale to meet any data collection needs in the future.

The final stream focuses on work being done by our German team.

Christian – Outage Timeline

Christian demonstrates the new outage timeline that I talked about earlier in the week.

Dustin – RRDtool export via ReST

Dustin shows a new feature that exposes collected data from the RRD files via ReST. This can allow for another integration point where collected data can easily be used by other applications.

Ronny – PRIS

The last demo was done by Ronny (presenting work that was also done by Dustin) on the Provisioning Integration Server (PRIS). As mentioned above, the ability to tightly integrate OpenNMS with provisioning systems is a key feature of the platform. Originally done to integrate with OCS Inventory NG, the system has been extended to allow for integration with pretty much any system.

Considering that these demonstrations were pretty much ad hoc, I was delighted to see how much was accomplished in just a week. It is one of the main reasons I look forward to Dev Jam every year.

We celebrated that evening with a trip to Republic.

Let’s just say the evening went a little downhill from there, but I did manage to make it back to the dorm.

2014 Dev Jam – Day 5

Friday, June 6th, 2014

Thursday turned out to be picture day. We had some people leaving a little early so we decided to get our group picture done in the morning. Recently the school added a new Goldy Gopher statue near our dorm, so it was the logical place for a photo.

It also worked out that Goldy would fit in a 3XL OpenNMS shirt (grin)

I even took the opportunity for one of them there “selfie” thingies:

Don’t expect to see many more, but I was told it would be “ironic” (grin).

There was some real work done as well. We are getting much closer to a 1.0 release of newts (http://www.newts.io), the NEW Time Series database built on Cassandra. The speed is pretty amazing, with sustained writes of 50K+ data points per second.

For testing we’ve been using some weather data that contains 1.2 billion data points, but even at 50K per second it takes six hours to import.

Note that this was done on Matt’s laptop and one Cassandra node. On server hardware it should be much faster, and Matt and Eric have worked very hard to make it linearly scalable: two nodes are twice as fast, four nodes are four times as fast, etc.

The whole Internet of Things paradigm requires the ability to manage massive amounts of time series data and we are getting close to making it a reality.

I am also dealing with the reality that I ate way too much pizza in the last 24 hours. Thanks to Chris Rodman and the good people at Papa John’s Pizza, we had a pizza feast:

(burp)

2014 Dev Jam – Day 4

Thursday, June 5th, 2014

First let me interrupt this blog post with a special announcement. A rather onerous security bug was discovered in OpenNMS that would allow any authenticated user to access pretty much any file on the system.

We felt it was bad enough to actually create a fix in the 1.10 branch as well as in the current stable, 1.12, so please consider upgrading at the earliest possibility.

Hat tip to Martin Laercher for reporting it.

Wednesday marked the halfway point in the week, and everyone seems to be in a good groove. With everyone able to work together in person, a lot of nifty things are getting done, including an upgrade to the latest version of Drools Expert.

The integration of OpenNMS with Drools allows for very powerful alarm correlation, and by migrating to Drools 6.0.1 it just got more powerful.

Wednesday also marked the day of the Twins game. For the past two years we’ve taken everyone to the Twins ballpark to watch a major league baseball game. It’s a beatiful place for baseball:

although they usually stuck us in far right field. Also, for the last two games the Twins played the Royals, and lost both times.

This year they put us in far left field:

and the Twins faced the Brewers. Since Milwaukee is close to Minneapolis there were a lot of Brewers fans in the stands, but the home team pulled it out for the win.

We got our name on the big board, too, which was cool, and Jeff was quick enough to catch a picture.

2014 Dev Jam – Day 3

Wednesday, June 4th, 2014

By Tuesday things are pretty much underway. People have divided up into teams and are busy working on various improvements to OpenNMS.

All of this is managed by Bamboo, our build and test system. We have a dashboard view of the status of each branch: green means good and red is bad.

Last year DJ was working on a Raspberry Pi controlled stoplight, and it is now complete and tied into Bamboo. Again, green means good and red means bad, but it also pulses when building.

Tuesday evening we decided to head to Mall of America for some blatant consumerism (and dinner). I hired a school bus to drive us there, and while a “short bus” might have been more appropriate for this crowd, we had too many people to fit.

As part of the evening’s festivities, we attended a Star Trek exhibition at the Mall.

It contained a number of costumes and props from the various Star Trek television series and movies. It wasn’t too amazing but it was fun, and folks seemed to enjoy it.

2014 Dev Jam – Day 2

Tuesday, June 3rd, 2014

The conference kicked off Monday morning. I did a general introduction and then handed it over to Matt (Brozowski), who in true un-conference fashion turned it over to the group.

One perennial topic with respect to OpenNMS has been to improve the GUI. One technology we are considering is AngularJS, so after the initial session a group of guys went across campus to one of the tech center classrooms so that Matt (Raykowski) could do a presentation on it.

But not all of the great things that come out of Dev Jam have to be on such a grand scale as “a new GUI”. Markus showed me a very useful feature called “Outage Timelines” that is already complete:

It allows to you quickly see the impact of outages in the last 24 hours. It will definitely be in the next major release and might even make it into the next 1.12 version.

For dinner Monday night I had Brasa cater once again. Even though I backed off on the size of the order, we tend to have a lot of leftovers so I figured it was best to have it early in the week. It was amazing as usual.

After dinner I had to walk some of it off, so I wandered around campus. The sky was threatening to storm, and the sunset against the Weisman Art Museum was beautiful.

So far the vibe at this years conference has been even more positive than usual. This is a week that reminds me why we work so hard the rest of the year. While OpenNMS isn’t a huge project, the people involved, myself exempted, are giants. It’s great to be able to spend time with them, no matter how brief.