Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Dean's Crispy Bloggette (avec Fromage)
#11
It just struck me that this should obviously be here in the beta section since I'm discussing future code stuff.
Dean Roddey
Software Geek Extraordinaire
Reply
#12
Oh, and last night I finally did a reasonably decent mix of my little music piece:

http://www.charmedquark.com/Web2/TmpAudio/Test5d.wma

Still isn't what it should be, but sounds more reasonable on speakers now.
Dean Roddey
Software Geek Extraordinaire
Reply
#13
Spent some of the day on the Z-Wave driver, but then got pushed off into looking into networking issues and definitely found one network level de-performance enhancement that go introduced in 2.0.14. It's not huge, but definitely isn't allowing optimal performance either. I'll get this fixed for the next release. I still have to dig into the original issue some more since I just found this in the process of looking for that.

Oh, and one more mix of my thingie:

http://www.charmedquark.com/Web2/TmpAudio/Test5e.wma

I'm slowly starting to get the hang of mixing and mastering. It's a bit of an art and I'm definitely just a novice, but this version is definitely much more polished and places the instruments more in their own 'sonic space' so that it's not so muddled.
Dean Roddey
Software Geek Extraordinaire
Reply
#14
So the release schedule has been changed a bit due to the fact that some networking issues were found, particularly related to wireless clients.

It turns out that wireless clients, which have a tendency to get into iffy connection situations, even when they are not far from the access point even, can cause really serious jam-ups in the CQC network layer (the ORB) of the servers being talked to by those clients, affecting other clients that otherwise are not having connectino problems. This is due to a serialization of the input into the server side ORB, meaning that one flakely client that causes the server to timeout waiting for it to get its act together will hold up everyone else.

This can cause seemingly random timeouts on the other clients, which seem to indicate that the server is locked up or something, but it's really not. This took some time to figure out out, and I wasted a fair amount of time looking through ways that the servers could lock up like that. As time went on, and it turns out that any of them might have this problem and that the greater the access to a given server the most likely it would happen, it became obvious it had to be in the server side networking. And eventually the light bulb went on. It was hard to figure out because the effects did not point towards the culprit, and the culprit was something that had worked just fine for years and only became a problem as people started using wireless clients more frequently of late.

In the process of fixing this, which involved changing the server side ORB architcture to dedicate a thread to spooling data back and forth to each client program, many other benefits have accrued. The speed of the system overall is much improved even in well behaved networks, because there is far more overlap of network traffic that before. The system is noticeably snappier and scales better with more clients being present. And, if a client is flakely, it affects that client and no one else.

There was also a smaller problem where a seemingly very innocuous change caused a delay in processing commands from the clients. It's not a huge issue, but it adds up and makes the system seem less responsive. This was introduced in 2.0.14. This has been fixed as well.

So, the bottom line is that because of the need to get the first fix out there for some existing customers who have a number of wireless clients, we are going to push forward with another 2.0.x release, instead of waiting for 2.1.

The only things that will be in this release are:

1. The fixes indicated above
2. An updated Z-Wave driver which is more or less the same as before but it supports the new style USB controllers and provides support for disabling or setting the speed of polling of each module. It has various other improvements internally, but in order to get this release out, they will not be fully realized in this next release. Some of those benefits are discussed earlier in this thread.
3. The Sage TCP/IP and updated AE900 drivers will be added
4. The CML BindLocal() socket method fix
5. I may get a couple other drivers into the build since that's fairly simple and safe to do.

But that's it. This driver needs to get out ASAP, and since very little has changed, that makes it as easy as possible to get it out reasonably quickly with a fairly short beta period since any problems should be easily localized and fixed because it would only be in the few areas that have changed.

I'm going to try to shoot for a beta release on monday.

* You will lose your existing Z-Wave module configuration, and will need to re-learn the network, because things work a lot differently in this new driver. But it will be well worth the trouble moving forward because of the many improvements.
Dean Roddey
Software Geek Extraordinaire
Reply
#15
I have installed 2.0.15 on my own system. It went suspciously well. Everything is humming along very nicely. One thing I've noticed is that it's so fast now that it sometimes gets interfaces up before the polling engine has gathered all the data again. I added a little pause in the interface loading to help with that, but now it's so fast that I may have to increase that a little more.

My interfaces, which aren't hugely image intensive, just appear pretty much instantly on clients. Even the media browsing ones take less than a second to pop up (slightly longer on the first access of a page after a restart, since the server has to load up the cover art images.) Now that I've browsed through it all, the pages are appearing all but instantly on a client machine.

The new Z-Wave driver seems to be doing very well also. You will lose your Z-Wave configuration, so be sure to write down the names of modules. I know it's a PITA, but it won't happen again any time soon. The benefits are worth it. I'm using the new Intermatic stick (stuck up on the wall with a two way sticky pad) and HA7 remote control.

Replication when smoothly. I set up my groups as they were before and renamed the modules and bumped them all up to a 4 second polling time. Watching the new 'polling load' field of the driver, it's not moved from zero so the polling rate could be decreased, but I'm not going to bother. Better to be conservative and a 4 second poll means an average latency of 2 seconds which is reasonable for my needs here. And there's also the issue out outgoing command latency to consider. Leaving more 'down time' for the driver to be idle means that outgoing commands can get through quicker and makes it more responsive.

The polling load indicates what percentage of the potentially pollable modules are currently queued up for polling (meaning that their poll time as arrived but they've not been polled yet.) If it's zero, that means that each unit is getting onto the queue and off again before another one is ready.

I've not finished getting the old style replication back into the driver yet. I figured I'd get it all working correctly otherwise, and get it installed on my system and make sure it's working in general while I go back and work on the old style replication.

So anyway, so far all is goodness and system responsiveness is way up, in my three system scenario here) even relative to the system before the little goober in 2.0.14 that slowed things down a bit. I'll be keeping an eye on it for the rest of the day and beating on it periodically. If it's still just as happy tomorrow, I'll put it up as a beta for a few folks to try out first to see how it goes. But I think that this will be a pretty painless release. That doesn't mean that someone won't find a thing or two of course.
Dean Roddey
Software Geek Extraordinaire
Reply
#16
All is still well. No glitches and it's still running fine. My 'blink the lights at 9' thing failed, because I spelled one of the modules wrong when I re-did the Z-Wave configuration, but otherwise it's all working as expected. I just brought up the IV on my work machine again and laughed out loud at how stupidly fast it loaded. The blink of an eye pretty much. So all is looking good still for a beta tomorrow.

So here's what I worked on in my late night fun time over the last couple days:

http://www.charmedquark.com/Web2/TmpAudio/Test9.wma

It's still a little snippet. I'm still working on small bits until I figure out all the tools. But I think I've about got it. This one came out quite well. I did one with all quite clean sounding instruments, so that I could really test out how good a final product I could get, and this one is very encouraging. It's late and I can't test this mix on the speakers, so it could be a little off relative to the headphones, but it sounds great in the headphones.

I've learned that there's a LOT more that comes after you finally get the parts played correctly. Well, I knew that there were. But then there's a LOT that comes after that also. I knew about the mixing stuff, but I didn't understand the whole mastering process, which is a whole different set of tools from the ones that you use to do the recording and mixing.

I was pretty bummed at first, since I just thought I wasn't going to ever get the kind of results I'd hoped for, but now I see that it can be done. It just took quite a lot of experiments with various settings at the steps along the way. The thing I'm missing now is a better drum synth, which I'm looking into now.

I actually did this little piece also as a potential sale, since a friend of mine does audio product (commercials and whatnot) and he wanted a piece for a commercial. I'm hoping that I can make a few bucks on the side, so that I can afford to get the other software toys I need to take it up another notch in quality. I figure that any little bit of bucks I make using the tools that I can invest in more tools.
Dean Roddey
Software Geek Extraordinaire
Reply
#17
Everything is still very happy this morning, so all is still well. No server stats are creeping up, and everything is still super-responsive. So I'm going to go back and get the old style Z-Wave replication working, and then get a beta drop up for later today.

BTW, while waiting for a build this morning, I remixed my little ditty above on the speakers, so it sounds a lot better now for regular speakers. Mixing in the headphones always ends up with a mix that is not right for speakers, but I have to do headphones only late at night. Anyway, give it another listen and it should sound much better on the computer speakers or desktop speakers.
Dean Roddey
Software Geek Extraordinaire
Reply
#18
I just posted two more tutorial videos, in the web server section.
Dean Roddey
Software Geek Extraordinaire
Reply
#19
Dean,

I just reviewed the videos for the Media Repo Config again and #2 eluded to a third video that would discuss player controls. Also, there is no discussion relative to the differences needed to start a video. I have a setup similar to yours with a Sony 777.

Just wondering if you still have #3 on the list to do and when you might be able to do it?
Larry in Denver
Reply
#20
Really, there's #3 through infinitely through all of them, since there's a fairly open ended amount of detail that could be discussed in each topic. But yeh, I'll get deeper into all of the areas once I've gotten all the basic coverage done.
Dean Roddey
Software Geek Extraordinaire
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)