So I decided earlier today to try a processor intensive action on my desktop and see how it fares. I have several long .ogg files (mixtapes created by a friend) that I want to play on my MP3 player. Sadly, my player doesn't support OGG Vorbis, so I have a need to convert them. Enter SoundConverter, a tool specifically designed to do such a thing.
Now, for a bit of description on my Desktop... It is a custom build. It has an Intel Motherboard, 8 core processor and 6 Gb of RAM. It also has 8Tb of attached storage and 60Gb(2x30 SSD drives) for OS and related programs. Needless to say, it is a beast.
I've never really tested it out. I realized the other day that I never really even needed to enter and change any of the BIOS settings(I looked through them, they were all set correctly from the factory). The Ubuntu Kernel Team and a great many of the other teams within the Ubuntu Project have wanted load testing that is effective for quite some time. It is a proposition that I have been investigating since I joined Canonical. It is also needless to say, it is a massive endeavor. How does one accomplish effective load testing on an OS and provide concrete data and results that indicate where possible problems may lie? I am still working on all of that. Now back to our situation at hand.
This machine is set up with a 1300Mb swap partition that it has never used. I set out, using the .ogg files to see what kind of pressure could be put on the machine before I see a stutter in my music that is also playing while the test is underway. (That being a major gripe of mine and many users as well)
I added 8 of the OGG files to the sound converter to be converted all at once. My load average at the time was around 1.8. I had Terminal open running htop, Chromium open with about 10 tabs of pages and my Rhythmbox app playing some DJ Chuckie. :)
All 8 cores went to 90+ percent when I started the conversion and moved between 85 and 95% all during the process, which lasted about 5 minutes. No swap usage and absolutely no skipping of the songs I was playing even when I moved to the Rhythmbox app and changed songs in the middle of the test. I also didn't notice any lag in window selection or loading like I used to on my older hardware. I'm very happy with that. :-)
This testing was conducted on the Ubuntu Maverick RC 64 bit. Great Job Ubuntu Project members!! :-D
Showing posts with label ubuntu-kernel. Show all posts
Showing posts with label ubuntu-kernel. Show all posts
Tuesday, October 5, 2010
Friday, September 3, 2010
Ubuntu Kernel Team Triage Summit
Since I've already blogged about it here I'll refrain from boring you with the detail there already.
What I will say is that I am very excited about this Summit. It is the first of its kind and I am really hoping that I haven't stunted it due to my failure to evangelize it enough before now. There simply hasn't been the time. Couple that with e-mail issues of my own creation and you have a recipe for failure.
I hope at least a few people hear about it and join in the fun. This should be an opportunity for folks who have very little exposure to the linux kernel and a desire to become more acquainted with the inner workings of linux to get some of the basic information they need in order to get much closer and provide some assistance to the team as we push to get eyes on the thousands of bugs we deal with daily.
I can't wait for Saturday Sept 11th. :)
~JFo
What I will say is that I am very excited about this Summit. It is the first of its kind and I am really hoping that I haven't stunted it due to my failure to evangelize it enough before now. There simply hasn't been the time. Couple that with e-mail issues of my own creation and you have a recipe for failure.
I hope at least a few people hear about it and join in the fun. This should be an opportunity for folks who have very little exposure to the linux kernel and a desire to become more acquainted with the inner workings of linux to get some of the basic information they need in order to get much closer and provide some assistance to the team as we push to get eyes on the thousands of bugs we deal with daily.
I can't wait for Saturday Sept 11th. :)
~JFo
Sunday, May 16, 2010
UDS Day 4
Day 4 began, as usual, with the Kernel Roundtable. Afterward, I attended the combined meeting to discuss the planned kernel version to be used in the Maverick release along with planned kernel versions across several other release topic branches. This was followed up by the Ubuntu Kernel Delta review where Leann worked with the other members of the team to determine what patches were ready to be removed due to upstreamed stable patches or superseded by other patches. After this was a discussion surrounding support for backporting the Maverick kernel to be supported on the server version of the OS. This is a brand new thing that we are trying so that server gets the latest server support in the LTS release. It should be very interesting.
Another nice lunch was had and then plenary sessions with the Debian Project Leader, as session entitled "What's this bit do? Ghosts of plumbing past, present and future" which was quite informative, followed by a group photo.
The evening's session that I was mainly interested in was the Launchpad/Kernel bug improvement session which I had requested earlier in the week to get the team together with launchpad to further define our requests with regard to items we'd like to see in the tool. My thanks go to Jono Lange for his input on this session from a Launchpad point of view and to Matt Zimmerman for helping to identify the needs of the specific requests as we discussed them.
Todo items as a result of the day's sessions:
* Too numerous to mention, but captured in gobby documents :-)
More tomorrow,
~JFo
Another nice lunch was had and then plenary sessions with the Debian Project Leader, as session entitled "What's this bit do? Ghosts of plumbing past, present and future" which was quite informative, followed by a group photo.
The evening's session that I was mainly interested in was the Launchpad/Kernel bug improvement session which I had requested earlier in the week to get the team together with launchpad to further define our requests with regard to items we'd like to see in the tool. My thanks go to Jono Lange for his input on this session from a Launchpad point of view and to Matt Zimmerman for helping to identify the needs of the specific requests as we discussed them.
Todo items as a result of the day's sessions:
* Too numerous to mention, but captured in gobby documents :-)
More tomorrow,
~JFo
Wednesday, May 12, 2010
UDS Day 3
Well, I feel pretty tired today. I am not yet convinced that this is the "UbuFlu" That I have heard about. I suspect it is more like a product of the pace and intensity of the sessions I have been a part of. I'll know more tomorrow. :-)
The Kernel Roundtable was first on the agenda as per normal. We discussed the things we all attended the day before and discussed work items that came about as a result.
The first session I attended today was about making Harvest more useable. This is a really useful tool that I am hoping to add to the items I use on a daily basis. I provided feedback where I could but this was only the second time I had seen it (a situation that was entirely my fault as it was on my TODO list).
Next was a private Kernel meeting to discuss a few topics followed by a very well attended session on multitouch support and its place in the next release. This was a packed house and I got a great deal out of the current support level versus what is expected to be ready for inclusion in Maverick. I hope they moved the follow on session to a bigger room. I was unable to attend the follow on.
Lunch was as normal, great selection with fantastic desserts. :-)
Plenaries covered the forward work planned for ARM followed by a quick look at Lucid workflow success and the Maverick Development plan as it stands now (more on this later). and lastly was a brief chat about translations (what a great group of people!) and an overview of the new font layout from the foundry. Very Exciting.
After Plenaries I was part of a session on Kernel Wiki 'gardening' which I hope will aid our community in finding information more easily as we move forward. Some relevant topics were, how to break out information on a per-release basis so that even as releases occur and information gets updated, our community will have access to relevant legacy information. This will also help us narrow the triage scope for specific subsystems community members are interested in and help us provide a better bug interaction. All big wins for the community as well as the bug reporters. I was also able to chat briefly about an idea that Steve Conklin had concerning the subsystems and triagers that focused specifically on subsystem bugs. The hope is to provide them with specific training from the triaging experts in these systems to allow us to not only grow their understanding of the specific sections of kernel work, but to also allow them to gain a foothold in kernel debug. The consensus was that this was a great idea and that the first iteration should be conducted electronically so that we can see the initial benefits without interrupting schedules of community members. This would also help us to live triage of specific bugs with an expert looking on and coaching. A side-effect that I hope we see is an expansion of the subsystem focused wiki pages that will come about as a result of the learning that occurs. I hope to use this as a model going forward for community based training that will eventually elevate current triagers into subsystem hackers and eventually kernel devs if they are interested. I guess you can sense my excitement. :-) More on this to come.
Next was a block of 2 hours devoted to kernel config which I opted to miss so that I could get a break and hopefully recover some energy for the final sessions of the day. The first was on Patch Review. Here again the Kernel Team is a bit different, but i took the time to try and understand how the other packages use it. The session was very informative. I now know exactly how the kernel patches deviate from this model and I understand more of the why. The last session I attended was apport and its possible use to gather information on installation failures. I always enjoy hearing the thought process that Colin Watson uses when determining what he needs to do to address an issue and this was a perfect opportunity to listen in. I was also mainly interested in what this might mean for kernel installation failures that might occur. i suspect I will be getting pinged by Colin should he encounter any of those in the testing. I look forward to it.
I've foregone dinner in the hope that rest will replenish me a bit more for tomorrow. We are 3 days down and 2 to go. :-)
Take aways from today's sessions:
*Read up on Harvest and pull the branch so that I can provide more feedback from a user perspective.
* Work with Andy to define the first needs of the wiki reorg
* Work the list of 10 for the initial foray into wiki updates
* Get arsenal scripts that pull bugs that have upstream SHA1 commits in them and generate a readable/clickable report for use in verifying that they either need to be or are in the latest kernel.
* develop a plan for the first 'triager summit' and announce intent to team list.
* Put the new Bug call plan in place and plan to start it one week from Monday the 17th of May.
* Work out a better bug report and have it update automatically through the day.
* Identify easily distinguishable bug tags to effectively break out the bugs as to what subsystem is affected (i.e. kernel-sound, kernel-suspend, etc.)
* Work up automated scripts to identify and tag as such above.
* roll triage wiki with links to the bug lists as broken out by the tags above.
* Determine best practice to expire bugs appropriately and work with Brad Figg to implement appropriately. (I have halted running the expire script until we have this down and i have reviewed all currently expired bugs)
* Work with Brad to identify the process arsenal scripts should use and document it.
* Rewrite arsenal scripts as needed due to the above.
* Identify Python training to attend over the cycle.
More tomorrow. :-)
~JFo
The Kernel Roundtable was first on the agenda as per normal. We discussed the things we all attended the day before and discussed work items that came about as a result.
The first session I attended today was about making Harvest more useable. This is a really useful tool that I am hoping to add to the items I use on a daily basis. I provided feedback where I could but this was only the second time I had seen it (a situation that was entirely my fault as it was on my TODO list).
Next was a private Kernel meeting to discuss a few topics followed by a very well attended session on multitouch support and its place in the next release. This was a packed house and I got a great deal out of the current support level versus what is expected to be ready for inclusion in Maverick. I hope they moved the follow on session to a bigger room. I was unable to attend the follow on.
Lunch was as normal, great selection with fantastic desserts. :-)
Plenaries covered the forward work planned for ARM followed by a quick look at Lucid workflow success and the Maverick Development plan as it stands now (more on this later). and lastly was a brief chat about translations (what a great group of people!) and an overview of the new font layout from the foundry. Very Exciting.
After Plenaries I was part of a session on Kernel Wiki 'gardening' which I hope will aid our community in finding information more easily as we move forward. Some relevant topics were, how to break out information on a per-release basis so that even as releases occur and information gets updated, our community will have access to relevant legacy information. This will also help us narrow the triage scope for specific subsystems community members are interested in and help us provide a better bug interaction. All big wins for the community as well as the bug reporters. I was also able to chat briefly about an idea that Steve Conklin had concerning the subsystems and triagers that focused specifically on subsystem bugs. The hope is to provide them with specific training from the triaging experts in these systems to allow us to not only grow their understanding of the specific sections of kernel work, but to also allow them to gain a foothold in kernel debug. The consensus was that this was a great idea and that the first iteration should be conducted electronically so that we can see the initial benefits without interrupting schedules of community members. This would also help us to live triage of specific bugs with an expert looking on and coaching. A side-effect that I hope we see is an expansion of the subsystem focused wiki pages that will come about as a result of the learning that occurs. I hope to use this as a model going forward for community based training that will eventually elevate current triagers into subsystem hackers and eventually kernel devs if they are interested. I guess you can sense my excitement. :-) More on this to come.
Next was a block of 2 hours devoted to kernel config which I opted to miss so that I could get a break and hopefully recover some energy for the final sessions of the day. The first was on Patch Review. Here again the Kernel Team is a bit different, but i took the time to try and understand how the other packages use it. The session was very informative. I now know exactly how the kernel patches deviate from this model and I understand more of the why. The last session I attended was apport and its possible use to gather information on installation failures. I always enjoy hearing the thought process that Colin Watson uses when determining what he needs to do to address an issue and this was a perfect opportunity to listen in. I was also mainly interested in what this might mean for kernel installation failures that might occur. i suspect I will be getting pinged by Colin should he encounter any of those in the testing. I look forward to it.
I've foregone dinner in the hope that rest will replenish me a bit more for tomorrow. We are 3 days down and 2 to go. :-)
Take aways from today's sessions:
*Read up on Harvest and pull the branch so that I can provide more feedback from a user perspective.
* Work with Andy to define the first needs of the wiki reorg
* Work the list of 10 for the initial foray into wiki updates
* Get arsenal scripts that pull bugs that have upstream SHA1 commits in them and generate a readable/clickable report for use in verifying that they either need to be or are in the latest kernel.
* develop a plan for the first 'triager summit' and announce intent to team list.
* Put the new Bug call plan in place and plan to start it one week from Monday the 17th of May.
* Work out a better bug report and have it update automatically through the day.
* Identify easily distinguishable bug tags to effectively break out the bugs as to what subsystem is affected (i.e. kernel-sound, kernel-suspend, etc.)
* Work up automated scripts to identify and tag as such above.
* roll triage wiki with links to the bug lists as broken out by the tags above.
* Determine best practice to expire bugs appropriately and work with Brad Figg to implement appropriately. (I have halted running the expire script until we have this down and i have reviewed all currently expired bugs)
* Work with Brad to identify the process arsenal scripts should use and document it.
* Rewrite arsenal scripts as needed due to the above.
* Identify Python training to attend over the cycle.
More tomorrow. :-)
~JFo
Tuesday, May 11, 2010
UDS Day 2
I feel much better today having slept 16 hours straight and only changing position once over the course of that time. :-)
Today started right in with sessions so I attended the Kernel Roundtable where we went over the sessions members attended from the day before. Steve has had an awesome idea to conduct a Kernel Triager summit for those triagers we have who are great at getting specific subsets of bugs to a state for work. I'll be proposing a session on this before the end of the week.
Next I attended the ARM toolchain session which was excellent, but a bit higher than I could reach with my current knowledge of ARM. i suspect this will be of more benefit as time goes on. :)
Next was an excellent session on btrfs, which I am looking forward to, wherein the conversation centered around what needs to be done to add this as an install option during installation of Maverick. The features of btrfs that were discussed make it sound imminently preferable to an EXT fs.
Next was a session about Ubuntu boot performance where there was an in-depth chat about what can be done to improve the boot performance in Maverick. There was some excellent information here on what currently causes delays in boot and what work is in progress to address this going forward. Very interesting stuff... if you are a geek like me.
Lunch was, again, delicious. It was followed by a riveting session on Diffamation which is an animated diff viewer. I think I shall be using it soon.
Next was a session on sounds which highlighted for me the methods used to change and identify sound changes for the next release. It was something that I was not familiar with that really helped me understand the process involved. The team is very focused on making the soundscape unique among the distributions.
I got a great deal out of the Tracking Regressions across Releases session. There are a great many things I need to look into as a result. This session was again hijacked by myself and the special needs nature of the kernel team and it's bugs. Fortunately, I recieved a ton of useful information in the way of apport and scripts that help highlight the relevant information in a bug from Matt Zimmerman. I plan to become fully conversant in apport before the end of the next cycle.
The final session of the day was on Permissive mounts of removable devices. There was a lively discussion surrounding expected behavior versus how it currently works.
Given the level of removable media issues I have seen reported, I felt like this was a great discussion to have.
Things to do as a result of today:
* Setup a Kernel Triage summit session to discuss the possibility of having this type of meeting.
* Meet briefly with JK to work out the HWE process for their bug going forward.
* Attend session with ogasawara, apw et Al to discuss further 'wiki gardening'.
* meet with BDMurray about greasemonkey, arsenal and apport.
* Reenable greasemonkey scripts and install arsenal.
* Read information send by Matt to start the apport learning. :-)
More tomorrow. :-)
~JFo
Today started right in with sessions so I attended the Kernel Roundtable where we went over the sessions members attended from the day before. Steve has had an awesome idea to conduct a Kernel Triager summit for those triagers we have who are great at getting specific subsets of bugs to a state for work. I'll be proposing a session on this before the end of the week.
Next I attended the ARM toolchain session which was excellent, but a bit higher than I could reach with my current knowledge of ARM. i suspect this will be of more benefit as time goes on. :)
Next was an excellent session on btrfs, which I am looking forward to, wherein the conversation centered around what needs to be done to add this as an install option during installation of Maverick. The features of btrfs that were discussed make it sound imminently preferable to an EXT fs.
Next was a session about Ubuntu boot performance where there was an in-depth chat about what can be done to improve the boot performance in Maverick. There was some excellent information here on what currently causes delays in boot and what work is in progress to address this going forward. Very interesting stuff... if you are a geek like me.
Lunch was, again, delicious. It was followed by a riveting session on Diffamation which is an animated diff viewer. I think I shall be using it soon.
Next was a session on sounds which highlighted for me the methods used to change and identify sound changes for the next release. It was something that I was not familiar with that really helped me understand the process involved. The team is very focused on making the soundscape unique among the distributions.
I got a great deal out of the Tracking Regressions across Releases session. There are a great many things I need to look into as a result. This session was again hijacked by myself and the special needs nature of the kernel team and it's bugs. Fortunately, I recieved a ton of useful information in the way of apport and scripts that help highlight the relevant information in a bug from Matt Zimmerman. I plan to become fully conversant in apport before the end of the next cycle.
The final session of the day was on Permissive mounts of removable devices. There was a lively discussion surrounding expected behavior versus how it currently works.
Given the level of removable media issues I have seen reported, I felt like this was a great discussion to have.
Things to do as a result of today:
* Setup a Kernel Triage summit session to discuss the possibility of having this type of meeting.
* Meet briefly with JK to work out the HWE process for their bug going forward.
* Attend session with ogasawara, apw et Al to discuss further 'wiki gardening'.
* meet with BDMurray about greasemonkey, arsenal and apport.
* Reenable greasemonkey scripts and install arsenal.
* Read information send by Matt to start the apport learning. :-)
More tomorrow. :-)
~JFo
Thursday, March 11, 2010
Who am I?
I am the Ubuntu Kernel Team Bug Triager. That basically means, if it is a bug on the kernel as shipped in a supported Ubuntu release, I manage it. This currently to the tune of 10,000+ bugs just on the linux(Ubuntu) package.
I work with the smartest team of people I have ever had the good fortune to come across, and I am thankful every day that they are so much smarter than me because they tend to see my errors before I make them public. ;-) If you have some time and you are interested in the inner workings of the Ubuntu kernel, feel free to come see us on FreeNode in the #ubuntu-kernel channel. There is a place for everyone.
Which brings me to another large part of my job, Ubuntu community involvement. If you are unfamiliar with how the Ubuntu community works, I can tell you that it is like one very big family all working together to make Ubuntu the best that it can possibly be for each release. Check out the pages here for more info. I can sure use the help managing the bug backlog for the kernel, so if you are interested in the kernel or in bug triage itself, chat with me on FreeNode and let me know you want to help.
That is all for now. Carry on.
~JFo
I work with the smartest team of people I have ever had the good fortune to come across, and I am thankful every day that they are so much smarter than me because they tend to see my errors before I make them public. ;-) If you have some time and you are interested in the inner workings of the Ubuntu kernel, feel free to come see us on FreeNode in the #ubuntu-kernel channel. There is a place for everyone.
Which brings me to another large part of my job, Ubuntu community involvement. If you are unfamiliar with how the Ubuntu community works, I can tell you that it is like one very big family all working together to make Ubuntu the best that it can possibly be for each release. Check out the pages here for more info. I can sure use the help managing the bug backlog for the kernel, so if you are interested in the kernel or in bug triage itself, chat with me on FreeNode and let me know you want to help.
That is all for now. Carry on.
~JFo
Subscribe to:
Posts (Atom)
