OpenNMS Update v1.34Nov 15, 2000, 06:39 (0 Talkback[s])
(Other stories by Shane O'Donnell)
Date: Tue, 14 Nov 2000 22:11:04 -0600 (CST)
Vol 1., Issue 34
In this week's installment... * Project Status + Starting to Stress Test + Starting to Stress Out + Coding Projects Underway * California, Here Steve Comes! * Shane Meets Maddog * The Wish List
Starting to Stress Test:
We're getting to a point with some of the underlying processes that we're able to start putting some things through their paces.
We've started with some initial stress testing of the events subsystem, including receiving traps, converting them to our internal XML-based Universal Event Identifier (UEI) format, pushing them through the event system (including a mapping phase for detailed event configuration, writing the event to the database, and invoking a Perl script as an automated action.
First, some more details on the test itself. We generated the traps first from the same box to a local trapd process (trap listener) via the loopback interface. We generated the traps in groups of 100, at a rate of roughly 10 traps/second. Handle it? No problem! And what's interesting is that it also accommodated writing these events to the database--which we were accessing across the network via JDBC. (Those of you thinking about how to scale this thing should be making mental notes about right now...)
The test environment: Bluebird 0.2.4 running on a P2-400 w/ 128MB (NT4) accessing a remote database server via JDBC running Postgres 7.0.2 on a P200 w/ 32MB (SuSE's stock 2.2.16-3 kernel).
We're continuing on, adding some additional code that will provide some more insight as to how things are being handled internally, where they may be bottlenecks, etc. And as we get more data, we'll be keeping you posted. Of course, as soon as we hit a production plateau, we have no doubt that we'll receive more info than we could ever have wanted if something doesn't perform well...
For those of you wanting to do some of your own stress testing now, I'd like to recommend you wait a couple days for simplicity's sake. We are just about to release SCM, which will automatically fire up some of the critical processes you need to run the "guts" of Bluebird, and as much as I hate to say it, the CVS build is still more focused on the UIs than the underlying components. Things are coming and are coming quickly. The Convergence is upon us!
Starting to Stress Out:
We've got an MP3 server in the corner responsible for bringing in da noise, and developers that stay in the office, code all night, and don't make it home to shower in the morning who are in charge of bringing in da funk.
We're on a collision course with Thanksgiving, and it appears that everything is going to be in place to meet that date, warts and all. But for those of you anxiously awaiting this release, keep in mind that much of this early functionality is coming at the expense of time spent with spouses/families/signficant others/alone, continually opting for pizza delivery over real food, and in a few cases, personal hygiene.
I can't say enough to laud the efforts that this team is putting forth. It's great to work with smart people, but smart people that are motivated and driven to succeed--that's a horse of a different color.
When we get through this initial push, have your travel agents send their brochures to the office, because some folks around here are in desperate need of vacations (and showers, but I already mentioned that...)
Coding Projects Underway:
* create.sql -- No changes this week.
* Events -- See the notes on the early stress-testing above.
* SCM -- One bug left to fix, and it was de-prioritized, since we know what the problem is and how to fix it. Changes and corresponding external release coming soon.
* OutageManager -- Vishwa's turned this over to Sowmya for final review, fixes, and certification. Should have something soon.
* RTC -- Jacinta is integrating the real-time console with the appropriate processes that drive it. Making good progress.
* icmpd/Perl -- Low priority. No update.
* Filters -- Working as advertised. All applause goes to Jason, with some saved for Weave's contribution of the Postgres stored procedures.
* Maji Prelim Work -- Haven't heard anything lately, but I know that Rick is plugging away.
* joeSNMP -- No changes.
* Operator Panels -- When I get bored, I screw around with the logo on the login panel. Feedback is appreciated.
* Admin Panels -- More minor bug fixes. Looks pretty solid.
* Documentation -- More updates available via the "devdocs" page
California, Here Steve Comes!
As promised and previously scheduled, Steve is hitting the road this week, making a swing by a couple different users' groups in the Bay Area.
If you are in the San Francisco/Sebastopol area, you can hook up with Steve either at the North Bay LUG or the BAADD meetings.
As always, details available at the Speaking Engagements link on the web site .
Shane Meets Maddog
In case you were thinking that this was about to be a story about my first experience with cheap wine, you are about 14 years too late, and definitely in the wrong forum (since I only talk about quaffing an occasional malty brew, here...)
Anyway, last week, with Luke and I in Austin and cordially hosted by the Austin LUG, I once again found myself in the same place as Jon "maddog" Hall, Executive Director of Linux International and Linux Evangelist #1 (We both did LinuxWorldExpo-Frankfurt and ALS). And when I headed to the airport to leave Austin, I had managed once again to miss meeting the man, the myth, the legend.
So alone and dejected, I hauled my /(carc)?ass/ to the plane and assumed my seat in 29D only to find, in 29F -- you guessed it -- a very attractive young lady! Oh yeah, Maddog was in 29A...
So I got the chance to introduce myself and talk a little bit about OpenNMS (I really don't have much else to talk about...), of which he seemed interested and noted that he'd be checking us out.
But wait, there's more to this story: As we talked, the flight attendant asked if we'd like to meet the pilot. So we head to the front of the plane, and before too long the conversation turns to peculiar dog behavior, Turkish prisons, and grown men in various states of undress. I could tell Maddog was kind of uncomfortable, so I politely excused us back to our seats where it was early enough that we were still able to take a nap/write some code before we were on the ground again.
I'll leave it to you to determine where the embellishment begins in this story, but if you've read this far and figure it out, you know now the two critical pieces of information: I got to meet and talk with Maddog and I'll go to just about any length for a cheap wine joke/Regular Expression/Airplane reference.
I gotta get more sleep...
The Wish List
As we close in on some critical dates, we need your help more than ever on some of these key project components. TIA!
* Documentation (both lightweight and detailed)
* Testing of the EUIs (once they are updated), with bugs reported to Bugzilla
* Add SNMPv3 functionality to joeSNMP
* Implement the event correlation spec.
* Consider the integration points for CIM/WBEM support
* Take a look at the ASN.1-to-XML conversion utilities that are available and let us know what you think... JavaCup parser for MIBs or ASN.1 to XML conversion.
I'm always looking to broaden my vocabulary, but I could have gone forever without introducing the phrase "hanging chad".