Forums :
Technical Support :
CAMB 2.11?
Message board moderation
Author | Message |
---|---|
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
I downloaded 2.10, but hadn\'t started any results yet. I decided to abort the 2.09 tasks, and when I reported them, 2.11 came down... So, what\'s new in 2.11? ![]() |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
I know I am a moderator here......But I feel we are having a real lack of communication here between the project and the users....we have apps 2.07-2,11 released with barely a response on the issues at hand....2.11 without a word....Cosmology has been a very stable Beta project and now has gone wonky with new releases that still have problems....can\'t we test these first before releasing them? I have sadly suspended cosmology on my 7 rigs,aborted any post-2.05 and await the confirmation of a working app. What is the problem of rolling back instead of using the community as guinea pigs? Many users have announced leaving the ship...is this approach good for the project? I am looking for a little re-direction here. Scott- I know you have worked hard trying to solve these problems this weekend...but why can\'t you communicate a little better based on the problems? |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
Cosmology has been a very stable Beta project and now has gone wonky with new releases that still have problems....can\'t we test these first before releasing them? Something to keep in mind is that the project is still Beta, thus we are \"testing\" the application... I posted this message probably within 5 minutes of 2.11 being available to download. I had made one pass of downloading new tasks and received 3 that were tagged as \"2.10\". I suspended project and aborted all of the 2.09 tasks, then resumed project and it requested more work along with the reporting of the aborted tasks. Since I had just received application 2.10, I wasn\'t expecting a new app version to come down along with it, but it did...and here we are... Anyway, the fact that the project is Beta is why I didn\'t like David Anderson pointing to this project as being \"out of line\" with credits. Part of the risk of running a Beta is that you can have things happen that you lose processing time. A perfect example of this is the fact that, at least with CAMB 2.09, \"CPU time\" as reported was going up very slowly compared to actual \"wall clock\" time... The credit per CPU time second would not have been reflective at all about the total time spent on the task. ![]() |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
Cosmology has been a very stable Beta project and now has gone wonky with new releases that still have problems....can\'t we test these first before releasing them? Ya Brian I am well aware of the \"beta\" status here but there are better ways of handling this than having most of the crunchers abandoning the project due to poor planning, I am sure problems weren\'t \"perceived\" as being an issue...but they are and mature projects test their apps befoe releasing( Most of us \"perceive\" this as being a production project), and most projects tell you of a new app before it is released.The last few days have been a \"nightmare\" indicative of an \"alpha\" project,not Beta. Most of the leaving crunchers were here in the first place because it \"was\" a stable project...you are totally missing my point.(as usual) You don\'t handle trying to go into being a production project this way. |
![]() Volunteer moderator Project administrator Project developer ![]() Send message Joined: 1 Apr 07 Posts: 662 Credit: 13,742 RAC: 0 |
At this point, there are problems which people are telling me exist, and I am trying my best to fix it as soon as possible. If that means we go through 5 versions in 3 days, then so be it. I just want to get it working ASAP. In the future, I will probably have a separate project up for testing since we had a bunch of problems with the test_camb application last time we tried it. In 2.11, I found a bug that seems to be the cause of the IO problems. This is the only thing that has changed. Scott Kruger Project Administrator, Cosmology@Home |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
At this point, there are problems which people are telling me exist, and I am trying my best to fix it as soon as possible. If that means we go through 5 versions in 3 days, then so be it. I just want to get it working ASAP. Thanks Scott ....the answer is worthy of a project heading toward production :) I had to raise these issues as no one else has and is \"very\" pertinent and sorry if I seem harsh.Please communicate better. |
Cameron Send message Joined: 2 Dec 07 Posts: 7 Credit: 145,565 RAC: 0 |
Hey that\'s nice to know. Keep working through the kinks. Maybe a brief message on the main page at your fixing problems with the client as you find/solve them and to expect new clients regularly over the coming days as you move to an overall improved client. I\'ve got a few day\'s solid work with some other BOINC Projects and will get some new cosmology work as as soon as I can. |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
I don\'t disagree with you. There is a better way... From where I sit, it looks like 2.06+ was not put through any kind of rigorous internal unit testing. Unit testing should\'ve caught the I/O thrashing. That said though, my Intel host was still running 2.08 and it did not have I/O thrashing. With my AMD host, I didn\'t see the trashing until I exited BOINC via right-click on the tray icon and selecting Exit. When I started up again, only then did I see the bouncing CPU times that were clearly due to an inordinant amount of time being spent on disk writes. Given the scenario above, it seems likely that there is some sort of \"trigger\" that causes the situation. Most likely that trigger is a hard shutdown of BOINC, be that an exit, a kill on the pid, or a reboot. My Intel host had none of that happen. It is a service installation and has not been rebooted, so BOINC has not been shut down... Of course, this could be a coincidence, but based on my observations, it seems like a good test scenario... ...mature projects test their apps befoe releasing Seems as though Scott has that intention, based on what he has said about offering a separate testing area... ( Most of us \"perceive\" this as being a production project), and most projects tell you of a new app before it is released.The last few days have been a \"nightmare\" indicative of an \"alpha\" project,not Beta. Most of the leaving crunchers were here in the first place because it \"was\" a stable project...you are totally missing my point.(as usual) Actually, I understood your point. The problem though is the expectation level that you have. The project is Beta. This means that problems should be expected. Should announcements have been made? Yes, they should\'ve. However, like they say, \"past performance not indicative of future results\". You got accustomed to \"smooth sailing\" and had a high level of confidence because of it. Same thing happened to Scott / the project team, IMO. Take-aways: - Project team needs to do a better job with unit testing and communication. - User base needs to have a better understanding that the project is Beta and not be lulled into a false sense of security. As always, IMO, YMMV, etc, etc, etc... ![]() |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
Thank-you Brian for you insiteful observations...howevever lulled or not people in a stable beta project expect better communication and output ...the reality is much crunching is lost and never should have been...lets chock this up as a learning experience and move on to bigger and better things....as long as the project learns from this experience. |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
Are there any kind of Change Management controls in use by the project? How about unit testing? Do you use any test scripts or automated testing tools? Would you consider having a small core group of testers for unit and/or QC/\"User Acceptance\" testing? Thanks... |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
howevever lulled or not people in a stable beta project expect better communication and output ...the reality is much crunching is lost and never should have been I do agree with the need for better communication. You and I however seem to disagree on the kind of expectations one should have with a Beta. You seem to have elevated this project to be Release Candidate (RC) status instead of Beta... I tend to consider all Beta projects as \"monitor carefully\", regardless of if they seem \"stable\" or not... This applies not only to the Cosmology@Home project, but all beta apps. It comes from my experience in project development... |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
howevever lulled or not people in a stable beta project expect better communication and output ...the reality is much crunching is lost and never should have been For example ....take a look at milkyway...alpha project... they announce all changes as they occur and post to threads with problems asap...a alpha project...cosmology -beta ought to exceed what is done in alpha and has not lately...and Brian its not \"my\" expectations that count......it is the community as a whole...as I have said from the beginning ( again you argue the point).I rest my case. |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
I agree... However, let me speculate on something. This week is either Midterm week or Spring Break week for most colleges and universities here in the United States... Perhaps a little bit of leniency could be granted...? |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
Perhaps you would feel different if you had 100 hosts attached...... |
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
Actually, no, I wouldn\'t... Now, as the Thread Owner, I respectfully ask you to cease the griping in this thread. If you want to gripe, please make another thread. |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
Griping? I think not....truth be told....If a break was anticipated the changes shouldn\'t have been made, And again you continue with weak arguements to sustain your feeble attempts to discredit me. I have your # Brian you posted that 2,08/2.09 had no problems when clearly you only had 2,05 in queue and said you had no problems....Not a noob as you said....worse you try to be Mr. know it all without substance.You will probably post to the nth degree of how wrong I am .....but the proof is in the pudding as they say...Anasas was being kind.He saw as I did you had no work in progress in question....he can back me up because I checked your results in progress when he posted.....give it up Brian...seems you only care about the last word in any thread and you wish to debate every post I make..Brian I am really tired of your shenanigans here. |
![]() Volunteer moderator Volunteer tester ![]() Send message Joined: 15 Jun 07 Posts: 345 Credit: 50,500 RAC: 0 |
Please take it to PM you two. There\'s no need for that in the open forums. I see Brian has started a new thread on 2.11, so anyone who still has problems can continue in that one. I\'m locking this one before it totally goes out of line. |