Seems the main page just got updated (Aug 9, 2004) with the following: (off a day, and there wasn't even a leap year involved!) (the upload first started working my first attempt after reading the message, so I know it's not from yesterday!)

August 8, 2004
The servers are back up as far as receiving results but not as far has distributing new workunits. We are simultaneously clearing out the old download directory and preparing to both it and the upload directory to the Snap Appliance, along with the DB. Allowing result uploads will ultimately speed both the deletion of old workunits and results and the granting of credit. Credit for the backlog of results will start being granted once we move to the Snap Appliance and turn on the hierarchical directory structure.
-----------------------------------------------
Ok, does the first part mean that schedulers are back up, so "ready to report" results are reported? I'm able to upload, but not contact the schedulers to "report" (the good old "no schedulers responded")

So when is the Snap box to be installed? (an estimate would be nice).

I think a fair translation of the latest news is "We'll we're doing part of the upload/report process, and won't be back up for other stuff until some time in the future, be it hours or days".

I know the "glass is half full" folks are getting ready to post "get job" messages to UCB and "flame AZ woody" messages to me, but please step back and think about what they said other than "we want to fix this", which should be clear even without a news post!

UCB - will the switch to Snap be done in a day, sometime this week, or next week? Give us a worst case estimate!

Hey Woody,

You know in some peoples mind I am sure some think it is
8/8.
Maybe I just confused people with my discription of "Epoch Time".

I am glad to see a Positive Post!! ;-)

Regards,

Rocky

> I am glad to see a Positive Post!! ;-)
>
> Regards,
>
> Rocky

No problem...

Some label me a a "troll" but, I do want to see BOINC move forward. As a Beta tester, I can say it was wayyyyyyyyyyyyy to early to go live! (I think it's safe to say that Dev was pushed by non technical issues to go live when they did)

I did find the "yesterday" post to be a bit lacking in "substance" (like, when can we expect to see WU's again?!)I find it odd that the news had the wrong date, but that's just me! In the same way, however, I have ~5200 credits from Seti and ~1500 credits from predictor. (and no new credit from seti, in what, 3 weeks?). Not a big farm, only 3 boxes....

If I wasn't hoping that UCB would get this working, I'd be gone!

I'll admit it, the "cheerleaders" got to me - they kept posting that "all is well", and there were those that claimed I was JQ or used multi logins!

The Boinc 4.x issue with climateprediction.net also raises an interstesting question. As Boinc is meant for a platform for "multi project DNC", seems BOINC needs to be "backwards compatible" as not all projects will support a "new BOINC" at the same time. (even for a major version number change) I hope the devs address (correct) backward compatability, as with only 3 BOINC projects, it's already resulted in posts in multiple forums. Imagine if there are 10 or 20 projects! (isn't that the goal of Boinc?)

As one of those you "proclaimed" a cheerleader, I want to remind you that I never was an "all is well" type person. I was a "try to understand" and "don't expect everything to be perfect" type person and always said that the release came a little early. I hated seeing people making all kinds of half cocked remarks without knowing anything about what BOINC is to be accomplishing or having any idea of what the SETI team is working with. While I have frequently agreed that the release was premature, I still defended Berkeley for making the decision for a variety of reasons. My main defense of the release still stands though. Regardless of when it was released with respect to readiness, can you honestly say that the Dev Team would have made NEARLY this much progress in the relatively short time that BOINC has been public? Many of the problems they have found were due to the severe overloading of the servers, some of which stress testing would have found, and some that stress testing wouldn't have found. To early or not, maybe it was for the best. I know it's been rocky, but that's just how things work sometimes and my goal was always to try to help people realize that instead of constantly coming down so hard on the overworked underpaid people at Berkeley with inadequate hardware resources at thier disposal. I also point out that while I got very irritated at your apparent attitude towards both myself and the project, I never labeled you as a "troll" or anything else of the like. This is the most positive and agreable post I have seen from you, and I hope to see more along these lines from you.

And you also make a very good point about the compatibility issues. If UCB distributed the software to all projects as soon as they released it, the timeframe of the incompatibility of certain projects could be kept to a minimum. Maybe there will be a clientside backwards compatibility (I.E.-V4 server works with V4 client but not V3 client, but V4 Client works with V3 and V4 server). Seems to me that combining those 2 methods would work out the best. It's definately a problem to consider and solve before too many projects get released though.

Chris

> No problem...
>
> Some label me a a "troll" but, I do want to see BOINC move forward. As a Beta
> tester, I can say it was wayyyyyyyyyyyyy to early to go live! (I think it's
> safe to say that Dev was pushed by non technical issues to go live when they
> did)
>
> I did find the "yesterday" post to be a bit lacking in "substance" (like, when
> can we expect to see WU's again?!)I find it odd that the news had the wrong
> date, but that's just me! In the same way, however, I have ~5200 credits from
> Seti and ~1500 credits from predictor. (and no new credit from seti, in what,
> 3 weeks?). Not a big farm, only 3 boxes....
>
> If I wasn't hoping that UCB would get this working, I'd be gone!
>
> I'll admit it, the "cheerleaders" got to me - they kept posting that "all is
> well", and there were those that claimed I was JQ or used multi logins!
>
> The Boinc 4.x issue with climateprediction.net also raises an interstesting
> question. As Boinc is meant for a platform for "multi project DNC", seems
> BOINC needs to be "backwards compatible" as not all projects will support a
> "new BOINC" at the same time. (even for a major version number change) I hope
> the devs address (correct) backward compatability, as with only 3 BOINC
> projects, it's already resulted in posts in multiple forums. Imagine if there
> are 10 or 20 projects! (isn't that the goal of Boinc?)
>
>
>

AZ..not to be flamed, a very postive and credible post.
I also found the new "news update" to be very interesting.
Now we can upload and report, but no guesstimate as to when
new work will be availabe. Nice post AZ.

Regards,

Xi3
<a> [/url]

> As one of those you "proclaimed" a cheerleader, I want to remind you that I
> never was an "all is well" type person. I was a "try to understand" and "don't
> expect everything to be perfect" type person and always said that the release
> came a little early. I hated seeing people making all kinds of half cocked
> remarks without knowing anything about what BOINC is to be accomplishing or
> having any idea of what the SETI team is working with.

The thing you missed in the picture with me, is that, in realtive terms, is that Boinc is not a large development project to me in terms of what I've done for a living for 20+ years. I'm not "guessing" about software development and efforts, but have been there myself. (on both sides of betas)


>While I have frequently
> agreed that the release was premature, I still defended Berkeley for making
> the decision for a variety of reasons. My main defense of the release still
> stands though. Regardless of when it was released with respect to readiness,
> can you honestly say that the Dev Team would have made NEARLY this much
> progress in the relatively short time that BOINC has been public?

No doubt about the fact that technically, it was a bad move. The Dev's reworked the bechmark code days prior to to 6/22, the logic for granting credit (2 to 3 WU's) was changed days before 6/22, the debug logic which meant that a 5 hr WU would take 30 hrs to crunch was removed days before 6/22, etc, etc, etc...

Basically, beta NEVER tested what when live on 6/22 (did I mention that the beta site was down for a few days prior to the "live" site?) Also, the 6/22 "live" anouncement didn't even make it to the beta for a few days. (I finally was able to download the new beta code the day it went live! No joke!)

Beta has 5000+ registered users, who'd been there for at least some of the 18 month beta.. Many weren't really doing much (like me) as things didn't seem to be progressing much... (I was with the beta for 7 months)

If, however, they sent out a "call" (they had the email for all beta testers, after all!), that "we want to stress/load test this, so starting x/x, please help!", they could have had a stress test without going live....... This could have been an easy step before "live" and would have provided a stress much like predictor was experiencing a few weeks back. It could have fixed amny a bottle nech before "live"

However, the technical folks were forced to release the software as "live" by a non-technical force (funding?) IMHO. That's why it wasn't really tested in beta, with all the new changes, and for loading.....


>Many of the
> problems they have found were due to the severe overloading of the servers,
> some of which stress testing would have found, and some that stress testing
> wouldn't have found.

Not really, if you understand the technology... A few PC's on the local LAN at UCB could have mimiced the load of all the folks on Seti/Boinc. By mimic, I don't mean crunching WU's, but connect/ask for work, upload, request more work... To stress test the "live" HW, they didn't even need the beta testers! (heck, we used stuff like this to stress test code back in the 80's - when each simulated user required an rs-232 cable to a terminal port)


> To early or not, maybe it was for the best. I know it's
> been rocky, but that's just how things work sometimes and my goal was always
> to try to help people realize that instead of constantly coming down so hard
> on the overworked underpaid people at Berkeley with inadequate hardware
> resources at thier disposal.

Aren't the folks at UCB on doing it for "credit" in a course? They are grad or under grad types, that may or not get paid, but are given college credit for their work, which they can include on resumes when they graduate! There are others , either employeed by UCB or hired as contactors, but in either case, trust me, it's not that they are earning minimum wage! Then there are those who bought the "open source" message and are coding/doing stat sites for free.

OK, so the seti Classic HW is 3x faster than that used by Seti/Boinc....

Classic has what, 200K-300k active users, while seti/boinc had 10k before things went south. Will the classic HW running seti/boinc handle 200k active users? (or 30k?)


>I also point out that while I got very irritated
> at your apparent attitude towards both myself and the project, I never labeled
> you as a "troll" or anything else of the like.

Why do you think my post was addressed toward you? I get hit with posts I do lump together as "cheerleaders" in my own head..... That lump changes in my own mind as I read posts.


>This is the most positive and
> agreable post I have seen from you, and I hope to see more along these lines
> from you.

You may want to review some of my earlier stuff, where I was getting shot down for no other reason than it was from "AZ woody"


>
> And you also make a very good point about the compatibility issues. If UCB
> distributed the software to all projects as soon as they released it, the
> timeframe of the incompatibility of certain projects could be kept to a
> minimum. Maybe there will be a clientside backwards compatibility (I.E.-V4
> server works with V4 client but not V3 client, but V4 Client works with V3 and
> V4 server). Seems to me that combining those 2 methods would work out the
> best. It's definately a problem to consider and solve before too many projects
> get released though.


This, to me, is a biggie..... After all, boinc is there for multiple projects, and 1+2 went Ok, but #3 threw a major wrench in the works! If I upgrade to Boinc 5.x for a specific project, climateprediction.net should still work even though it's on 4.x, as should set on 3.x

Let me apogize to anyone that I may have flamed in this thread.

We're all just trying to make this thing work, and I do come off abrasive at times!

(BTW, ET called me today, and asked if he could come over for a beer...) :)