It starts working, and next we know is "Boinc 4 tomorrow", and the new splitters now have the wrong URL!

Why can't UCB just say "ok it works, and the crunchers have been angry, so how about we just keep it running and take a break for a few days before we change stuff?

What happened is classic (well, not classic, but New Coke), in that the new splitters seem to be messing up the URL. They "announced" having 3 splitters back in late june or early july, so this isn't a new change!

So now, a few days after month old credits started to be granted, we find that new work is in limbo, and 4.x will be required to get work tomorrow!

Pope Dave must be ruling this with an iron hand for some reason!

Well for one thing the splitters weren't up and running, so within the next couple of days there would be no work to hand out.

The new splitters had to be used to handle the new directory fanout scheme adopted two weeks ago to get around a file system performance problem that was slowing all the backend services down.

We can't have it both ways, the system as it stands cannot handle the load of classic, so as performance bottlenecks are identified, they are fixed. This issue couldn't have been found in alpha, as alpha uses a different workunit generation tool against a known workunit. Beta would have been able to find it but only after a few days, or possibly weeks.

Just because we announced that we had three machines allocated for doing splitting work doesn't mean the splitters don't get updated when a major change occurs in the system.

Hell, the validator wasn't fixed until Monday, the assimilator and file_deleter weren't fixed until yesterday. Today the splitters came back online, they didn't report any errors and so this bug slipped through the cracks.

Progress is being made, but software issues are generally found after the hardware issues are resolved, and last week was all hardware issues.

----- Rom
BOINC Development Team, U.C. Berkeley

This should be interesting. Back row seat, beer's cold, popcorns popping, just can't wait for the show to begin.

---

Account frozen...

> What happened is classic (well, not classic, but New Coke), in that the new
> splitters seem to be messing up the URL. They "announced" having 3 splitters
> back in late june or early july, so this isn't a new change!

[rant mode on]

For some wazoo coder you make yourself out to be, you seem to not understand the various checkin notes. Hell its spelled out in black in white which files are touched and what was changed. Now if you were really as wazoo as you say you are you would be able to put two and two together and make a reasonable statement.

David 6 Aug 2004
- Changed server software to use the hierarchical directory mechanism.
This means using dir_hier_path() to generate the pathname
of an input or output file,
rather than just concatenating directory and filename.
Also (in process_wu_template()) use dir_hier_url()
to generate the appropriate URL
- Add a scheme for transitioning from flat to hierarchical download dir.
Can have a element in config.xml,
giving the path of an old, flat dir.
file_deleter will look here if it can't find file in main download dir.
- Pass around SCHED_CONFIG in a few interfaces
(e.g. create_work()) in order to access uldl_dir_fanout

html/inc/
profile.inc
lib/
util.C,h
sched/
file_deleter.C
file_upload_handler.C
make_work.C
sched_config.C,h
validate_util.C
wu_check.C
tools/
Makefile.am
backend_lib.C,h
create_work.C
dir_hier_move.C
dir_hier_path.C

So lets see here, validate_util.C, check, so the validators are affected.
Oh, file_deleter.C was modified as well, I guess that means the file_deleter was modified as well. create_work and make_work which are used by the alpha project.

What a stupid ass comment to make "They "announced" having 3 splitters back in late june or early july, so this isn't a new change", so according to the above checkin notes, the splitters haven't been running since the 7th, as they wouldn't have been compatible with the new design.

I would have expected more from a self described coder at the top of his game.

[rant mode off]

As for the peanut gallery, you cannot be more frustrated than I am.

----- Rom
BOINC Development Team, U.C. Berkeley

Unfortunately, there is no Internet RFC that describes NCTP.

We could send AZ Woody an NCTP client and maybe he'd figure something out.

> > What happened is classic (well, not classic, but New Coke), in that the
> new
> > splitters seem to be messing up the URL. They "announced" having 3
> splitters
> > back in late june or early july, so this isn't a new change!
>
> [rant mode on]
>
> For some wazoo coder you make yourself out to be, you seem to not understand
> the various checkin notes. Hell its spelled out in black in white which files
> are touched and what was changed. Now if you were really as wazoo as you say
> you are you would be able to put two and two together and make a reasonable
> statement.
>
> David 6 Aug 2004
> - Changed server software to use the hierarchical directory mechanism.
> This means using dir_hier_path() to generate the pathname
> of an input or output file,
> rather than just concatenating directory and filename.
> Also (in process_wu_template()) use dir_hier_url()
> to generate the appropriate URL
> - Add a scheme for transitioning from flat to hierarchical download dir.
> Can have a element in config.xml,
> giving the path of an old, flat dir.
> file_deleter will look here if it can't find file in main download
> dir.
> - Pass around SCHED_CONFIG in a few interfaces
> (e.g. create_work()) in order to access uldl_dir_fanout
>
> html/inc/
> profile.inc
> lib/
> util.C,h
> sched/
> file_deleter.C
> file_upload_handler.C
> make_work.C
> sched_config.C,h
> validate_util.C
> wu_check.C
> tools/
> Makefile.am
> backend_lib.C,h
> create_work.C
> dir_hier_move.C
> dir_hier_path.C
>
> So lets see here, validate_util.C, check, so the validators are affected.
> Oh, file_deleter.C was modified as well, I guess that means the file_deleter
> was modified as well. create_work and make_work which are used by the alpha
> project.
>
> What a stupid ass comment to make "They "announced" having 3 splitters back in
> late june or early july, so this isn't a new change", so according to the
> above checkin notes, the splitters haven't been running since the 7th, as they
> wouldn't have been compatible with the new design.
>
> I would have expected more from a self described coder at the top of his
> game.
>
> [rant mode off]
>
> As for the peanut gallery, you cannot be more frustrated than I am.
>
> ----- Rom
> BOINC Development Team, U.C. Berkeley
> <a> href="http://www.boinc.dk/index.php?page=user_statistics&project=sah&userid=85465">
>

> > What happened is classic (well, not classic, but New Coke), in that the
> new
> > splitters seem to be messing up the URL. They "announced" having 3
> splitters
> > back in late june or early july, so this isn't a new change!
>
> [rant mode on]
>
> For some wazoo coder you make yourself out to be, you seem to not understand
> the various checkin notes. Hell its spelled out in black in white which files
> are touched and what was changed. Now if you were really as wazoo as you say
> you are you would be able to put two and two together and make a reasonable
> statement.
>
> David 6 Aug 2004
> - Changed server software to use the hierarchical directory mechanism.
> This means using dir_hier_path() to generate the pathname
> of an input or output file,
> rather than just concatenating directory and filename.
> Also (in process_wu_template()) use dir_hier_url()
> to generate the appropriate URL
> - Add a scheme for transitioning from flat to hierarchical download dir.
> Can have a element in config.xml,
> giving the path of an old, flat dir.
> file_deleter will look here if it can't find file in main download
> dir.
> - Pass around SCHED_CONFIG in a few interfaces
> (e.g. create_work()) in order to access uldl_dir_fanout
>
> html/inc/
> profile.inc
> lib/
> util.C,h
> sched/
> file_deleter.C
> file_upload_handler.C
> make_work.C
> sched_config.C,h
> validate_util.C
> wu_check.C
> tools/
> Makefile.am
> backend_lib.C,h
> create_work.C
> dir_hier_move.C
> dir_hier_path.C
>
> So lets see here, validate_util.C, check, so the validators are affected.
> Oh, file_deleter.C was modified as well, I guess that means the file_deleter
> was modified as well. create_work and make_work which are used by the alpha
> project.
>
> What a stupid ass comment to make "They "announced" having 3 splitters back in
> late june or early july, so this isn't a new change", so according to the
> above checkin notes, the splitters haven't been running since the 7th, as they
> wouldn't have been compatible with the new design.
>
> I would have expected more from a self described coder at the top of his
> game.
>
> [rant mode off]
>
> As for the peanut gallery, you cannot be more frustrated than I am.
>
> ----- Rom
> BOINC Development Team, U.C. Berkeley
> <a> href="http://www.boinc.dk/index.php?page=user_statistics&project=sah&userid=85465">
>

No matter what our 'EXPERT' Woodrow has to say Rom, you guys and gals are doing a bang up job. I know it is hard but try to ignore the loud mouth nitwit. And Woodrow you have Rom poed now, so you should slink off to predictor cause they run thiers more to your liking....... Woodrow I just remembering an old Army saying, When confronted by an impossible situation, adapt and overcome, never hurts to be versatile.........Have a nice day......

This 'SPACE' for Rent.
It is not my fault, I am a cranky Oldman. ;o(

M7

> No matter what our 'EXPERT' Woodrow has to say Rom, you guys and gals are
> doing a bang up job. I know it is hard but try to ignore the loud mouth
> nitwit. And Woodrow you have Rom poed now, so you should slink off to
> predictor cause they run thiers more to your liking....... Woodrow I just
> remembering an old Army saying, When confronted by an impossible situation,
> adapt and overcome, never hurts to be versatile.........Have a nice day......
>



"Ditto"

>
> [rant mode off]
>
> As for the peanut gallery, you cannot be more frustrated than I am.
>
> ----- Rom
> BOINC Development Team, U.C. Berkeley
>
>


From us here in the peanut gallery, thanks for all your hard work.
We are hanging in while you kill these little bugs. :)

> > What happened is classic (well, not classic, but New Coke), in that the
> new
> > splitters seem to be messing up the URL. They "announced" having 3
> splitters
> > back in late june or early july, so this isn't a new change!
>
> [rant mode on]
>
> For some wazoo coder you make yourself out to be, you seem to not understand
> the various checkin notes. Hell its spelled out in black in white which files
> are touched and what was changed. Now if you were really as wazoo as you say
> you are you would be able to put two and two together and make a reasonable
> statement.
>
> So lets see here, validate_util.C, check, so the validators are affected.
> Oh, file_deleter.C was modified as well, I guess that means the file_deleter
> was modified as well. create_work and make_work which are used by the alpha
> project.
>
> What a stupid ass comment to make "They "announced" having 3 splitters back in
> late june or early july, so this isn't a new change", so according to the
> above checkin notes, the splitters haven't been running since the 7th, as they
> wouldn't have been compatible with the new design.
>
> I would have expected more from a self described coder at the top of his
> game.
>
> [rant mode off]
>
> As for the peanut gallery, you cannot be more frustrated than I am.
>
> ----- Rom
> BOINC Development Team, U.C. Berkeley

From the outside I can see where he is frustrated but with the size and scope of this project that you are trying to build you guys have done an awsome job. If we in the community here can help you guys in Berkeley out let us know and I sure you will be overwelhmed with positive reponses!

Keep up the good work!



Join the World Inc!

Rom,

> As for the peanut gallery, you cannot be more frustrated than I am.
>
> ----- Rom

We are really, for the most part, on your side. Honest! :)

I do hope that we continue to make progress and you are absolutely right that you have to shoot the bugs in order of identification and isolation. And with performance issues you have to find them one by one.

I am constantly amazed by the number of posters here that have had a lifetime of perfect projects where everything worked as expected and they anticipated all problems and they were omnipotent.

In my decade of doing software I worked on about 7 major projects and none of them were what I would call flawless. But, then again, I am not omnipotent ... :)

Keep the chin up, nose to the grindstone, hand on the keyboard, eye on the horizon ...

Man thats got to be uncomfortable!

Woody is a complete loser and one of the biggest tools I've ever run into on the internet. He's one of the few people I've "met" online that's so pathetic and warped that he has to take ANYTHING positive and turn it into a negaitve. If you go over to the Predictor board, someone will say something like "Predictor is running great - thanks developers!" and Woody, right on cue, will post "...UNLIKE SETI!!!! AHAHAHHAHA IM CLEVER!!!!!!"

What a sad little man :( I'm just a simple 23 year old graduate student, and I live a very simple life. However, even from my vatange point, I pity Woody as a human being.

Of course now he's just going to make a retort based on his "wit" and "expertise." I look forward to reading it. It will be worth a laugh :)

Anyway, Rom, ignore the toolboxes like Woody and his posse of fellow sleazes. Seti is doing fine, and thanks for putting all the hard work into things! RA RA RA!

Proud to be a cheerleader.

> This should be interesting. Back row seat, beer's cold, popcorns popping,
> just can't wait for the show to begin.
The main problem with this show is that it is a remake of the show that they keep rerunning on the other channel... and I'm out of beer!
Classic S@H is my fallback so I still feel that I am contributing to the cause one way or another.

> Pope Dave must be ruling this with an iron hand for some reason!


AZ Woody AKA conceited Pea-cock God

"I am Pope of SETI@home,

I am better than all Berkeley,

Anderson newbie,

vote AZ Woody for SETI@home new director!!!"
S@h Berkeley's Staff Friends Club © member

I can't help but think that the Woody's point has been missed here. Yes, there have been hardware issues, and yes, software problems have to be addressed as they come to light, but surely with the recent history of SETI/BOINC now is not the time to introduce a major client change?

If, as Rom rightly suggests, there have been software issues that could not have been found in testing, then surely now is the time for some stability so that remaining issues can be worked through.

I don't like Woody's style, mostly, but I do think a delay of a couple of weeks while the dust settles would have been prudent.


Giskard - the first telepathic robot.

I have and will say it again. the SETI team has done an excellent job with keeping this project up, and ironing out any bugs. Now AZ Woody, just shut up. ok? shut up. so there have been problems with the project. The team is trying to fix these problems ASAP, without too many problems. Every post i've seen of you has been ill towards the team, BOINC, SETI etc. If you have nothing positive or encouraging to say, don't say it. we don't need it. we don't need you. I respect your opinion and such, but we just don't need it.

MikeW, we need to release BOINC 4 soon to keep up with other DC projects, such as CPDN, so we don't have 2 clients conflicting with each other, as the 2 major versions (3 and 4) are incompatible with each other.

sorry to vent like this, but it slightly pisses me off. Perhaps take a hint from what we are saying.

-Fox




S@h Berkeley's Staff Friends Club © member

[/url]

> I can't help but think that the Woody's point has been missed here. Yes, there
> have been hardware issues, and yes, software problems have to be addressed as
> they come to light, but surely with the recent history of SETI/BOINC now is
> not the time to introduce a major client change?
>
> If, as Rom rightly suggests, there have been software issues that could not
> have been found in testing, then surely now is the time for some stability so
> that remaining issues can be worked through.
>
> I don't like Woody's style, mostly, but I do think a delay of a couple of
> weeks while the dust settles would have been prudent.
>
>
> Giskard - the first telepathic robot.
>

Mike:

Is this not the classic case of keeping up with the Jones? CPN has now gone to 4.05, of which I am running at home. The other projects, which I run at work, on slower computers, run SETI and Predictor. In order to stay with the elite upgraders and keep a user base, SETI and Predictor have to go to 4.05 soon.

Just my thoughts.

WCK

I've seen it all before. Some muckhead chimes in when it isn't necessary insulting the UCB crew for no particular reason. Usually they are wrong and even when they have a valid point (unlike the current situation) they go about it all wrong.

In the 5 years that I have participated in this porject there have been moments where I have equated UCB with "yUo Can' BE serious," but not once did I ever voice that opinion because while waiting, reading everything, and letting the real dunderheads (I'm only pretending to be one) stick their necks out for me.

That being said, I think UCB has done a creditable job of putting this distribution project together. Yes, there are always the backseat coders that think they know how to do it better, but they aren't doing it while UCB is.

The nitpickers need to make a choice:

1) Shut up and deal (my favorite cribbage quote)
2) Apply for a job on the crew (not likely unless you really are hot snot)
3) Quit the project (that includes all BOINC projects)

The third would be my recommendation to them. As any businessman can tell you, if you lose 10% of your clients, so what? If you lose 90% of your clients because of the other 10% you're in deep doo. I'd rather not see BOINC suffer because of people who may not be entirely stump-dumb, but definitely require an upgrade to thier social-censor function.

Closing: Yes, Virginia, there is an upgrade path.

-nosig.toodangold-

> I don't like Woody's style, mostly, but I do think a delay of a couple of
> weeks while the dust settles would have been prudent.

The issue though is that we now have 3 projects using BOINC and two are using one version and the other one is using a different version. With some of us sitting on the sidelines with BOINC and back to Classic, well, it would be nice to have all three projects using the same baseline so that the developers of all three projects can pool the resources and we do not have efforts that are conflicting.

I have worked on projects with two baselines in flight and almost any amount of pain is worth it to get down to only one baseline. With the major step upgrade it gets all of the BOINC Software versions alligned and we can now have "dot" upgrades that you can do at your discression (unless there is a show-stopper bug forcing a major version change).

Heck, it is possible that I will fire up one computer again with BOINC to have it do all three projects ...

One other point is that with all three in the same version we can get work from all three projects and maybe relieve some of the pressure on the SETI Servers.

Just some thoughts...

OK, I'll be the "muckhead". All I want to do is contribute to the project by crunching work units. I don't have time to do research in the message boards to piece together a plan of action.

Bogus Work Units? Which one's are those? OK, I guess they can't tell me which one's are bogus in the news message. I have to search for information about them in the message boards. Why can't I report some 30 WU's I've crunched on one of my computers that already has over 5000 credits (computer ID 8413)? I go to an identical machine using the same phone line and it works fine (yes, I have tried "update" and yes, both computers have the same amount of memory, processors, winXP with the same updates, etc)

Why isn't there an "upgrade procedure (plan) to version 4? I see a bunch of "well, you should run your cache out first", "you shouldn't have problems if you exit then upgrade", "you should detach","you shouldn't detach", "don't reset", "change your preferences to run your work out", you don't have to run your work out ........".

I would expect the "upgrade plan" to be released along with the upgrade.

Until then, I don't see a deep need to worry about it one way or the other.

> Why isn't there an "upgrade procedure (plan) to version 4? I see a bunch of
> "well, you should run your cache out first", "you shouldn't have problems if
> you exit then upgrade", "you should detach","you shouldn't detach", "don't
> reset", "change your preferences to run your work out", you don't have to run
> your work out ........".
>