what's wrong with raising/removing the limits for the outage...there is time to affect a change and for those of us that wish to have a cache of seti work to build that cache before relocation...

---

what's wrong with raising/removing the limits for the outage...there is time to affect a change and for those of us that wish to have a cache of seti work to build that cache before relocation...

Lionel.......
Nobody's listening to that argument.
And nobody's gonna.
With the project running fine from hand to mouth, nobody wants to hear about raising the limits anymore.

Our contribution has outweighed our opinion on the matter.

I, no matter how much I contribute, don't have a say these days in the decision.

Not even from the kitties.


That being said, I would truly sleep better if I had back my cache of some 70,000 WUs. I never turned one in late. Other than server conniptions.

The kitties always turned their work in on time. Always.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

what's wrong with raising/removing the limits for the outage...there is time to affect a change and for those of us that wish to have a cache of seti work to build that cache before relocation...

As long as you and all the other high volume crunchers promise;

1 - Not to all try and download a full cache at the same time.

2 - Not to all try and upload 3 days worth of your results and demand new work as soon as the project comes back online!

Please be sensible a massive download and upload at this time makes no sense!

---

what are you worried about ... that I'll get closer to you ... yeah, in about a year ... btw, have you noticed what's coming up underneath you ... :))

cheers

Who, what???
Morten and juan have blown by me and the kitties.
Nobody in sight after that.

Don´t worry, with the new 680's the kitties will retake their place soon.

---

what are you worried about ... that I'll get closer to you ... yeah, in about a year ... btw, have you noticed what's coming up underneath you ... :))

cheers

Who, what???
Morten and juan have blown by me and the kitties.
Nobody in sight after that.

Don´t worry, with the new 680's the kitties will retake their place soon.

My comment was to T.A., not to Mark but that's ok...

There are 2 coming up fast underneath me and should over take me in circa 35 days...and there are another 2 coming up kind of fast under T.A but I can't remember their run rates, so based on circumstances remaining fairly constant he will eventually drop 4 slots ... that's all :))

Mark, you may be right in that they are not interested in listening to their user base. The current situation can't last for too long. I am referring to the TCP fix. The fix improves link performance but, as we add more machines with faster processing CPUs and GPUs, we are going to hit the same issue again with congestion. We have already had the first wake up call. The second is on its way.

cheers mate ...

---

I was mainly a mainframe. I remember starting with COBOL back in 1969 when there were 3 different versions depending on the operating system (TOS - Tape Operating System, DOS - Disk Operating System & OS - Basic Operating system) you used at the time. I've seen it grow from that to the current version as of 4 years ago -- 44 years and it will not die regardless of what the PC world believes.

Back between 1988 & 1990, I did code and testing using COBOL , Assembler & FORTRAN on a HP portable that emulated a mainframe MVS/CICS system that weight about 20 pounds and had a red screen. Awfully hard to read anything on that bad boy, so we hacked it and changed to screen to black with yellow characters.

Had to monitor the refrigeration to keep it running. Had an alarm at my desk if things got too toasty. Hard drives as big as a Yugo. I think they had about a half a dozen platters in them. They were about the size of a small top loading washing machine, actually.

I had to change them out once in a while.......'bout like changing the litter box.

And about as big. They were about 24 inches wide and 18 inches tall, weighed about 20 pounds or more. Had this big handle on the top that you twisted to engage the thing into the drive. And talk about spin-up? You could hear those things winding up for miles around...LOL.

Our professor had an old reel to reel behind his desk i dont remember what system it was from anymore and he used to answer all complaints about storage space, speed, and most other complains about the systems we were using, by pointing to it :)

cheers mate ...

You too.

I engage in the discussion because, as a senior member of the crew, I must.
Not everybody agrees with my opinions.

I sometimes have insides to what is happening, but not always.

For example, I was totally left out of the discussion about the relocation.

I asked several questions. But none were answered.

So, I am just another user, as are you.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

I'm trying to anticipate what will look different from our end. Mark has already commented in the News thread that our current Cricket graph won't apply, and that will be a loss. Don't know if another will be appropriate, or whether our traffic will be mixed with other traffic. That graph has been the most useful indicator of project status for me. Wonder if the IP addresses will change? I'll probably neuter my hosts file during the downtime to avoid a surprise when downloads resume.

We're all hoping for more bandwidth, but I think that's still a maybe. The improved power, a/c and UPS coverage should increase reliability and server life, and the 24/7 coverage should help a lot, although the level of service is not clear. Anything else likely to look different?

---

Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.

cheers mate ...

You too.

I engage in the discussion because, as a senior member of the crew, I must.
Not everybody agrees with my opinions.

I sometimes have insides to what is happening, but not always.

For example, I was totally left out of the discussion about the relocation.

I asked several questions. But none were answered.

So, I am just another user, as are you.

FWIW as far as I can tell they kept that completely internal.

I still think raising the limits wouldn't achieve much at this point. We are maxxed out and have been for a while. work is going out as fast as the infrastructure supports. If the limits were raised, the high end machines would probably have some more work. at the same time there is only so much cake - your share gets bigger, other shares get smaller. It may distribute over so many machines that it's not very noticeable but stil. Only so much cake.
Whether the throughput would be higher is anyones guess.

Scarecrow used to have that very enlightening graph about what part of the work was done by what RAC groups ;)

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

I was mainly a mainframe. I remember starting with COBOL back in 1969 when there were 3 different versions depending on the operating system (TOS - Tape Operating System, DOS - Disk Operating System & OS - Basic Operating system) you used at the time. I've seen it grow from that to the current version as of 4 years ago -- 44 years and it will not die regardless of what the PC world believes.

Back between 1988 & 1990, I did code and testing using COBOL , Assembler & FORTRAN on a HP portable that emulated a mainframe MVS/CICS system that weight about 20 pounds and had a red screen. Awfully hard to read anything on that bad boy, so we hacked it and changed to screen to black with yellow characters.

Anybody here work on a Basic4 system?
I programmed one 30 some years ago.
Idiot software engineer just refused to give some folks the printer output they desired. Once the word got out that I could reverse engineer his crap.......
Let's just say the traffic to my desk did not let up much.

Had to monitor the refrigeration to keep it running. Had an alarm at my desk if things got too toasty. Hard drives as big as a Yugo. I think they had about a half a dozen platters in them. They were about the size of a small top loading washing machine, actually.

I had to change them out once in a while.......'bout like changing the litter box.

And about as big. They were about 24 inches wide and 18 inches tall, weighed about 20 pounds or more. Had this big handle on the top that you twisted to engage the thing into the drive. And talk about spin-up? You could hear those things winding up for miles around...LOL.

And whilst we are on the subject, anybody ever change the ribbon on a Centronix band printer? That thing wound up like a banshee too......and could spit out a full box of paper in a few minutes.

Those were the heady days of IT, back then it was called either Information Systems (IS) or Information Management systems (MIS). My first system was an IBM360 mod 30, Disk Operating System that had 2 banks of 2311 disk drives, 10 each; and 2 banks of 2411 tape drives 8 each. Between all of the drives spooling up and the 4 1403 printers banging away with the doors open, I almost forgot the 6 Magnetic Ink Character Readers/sorters (MICR); some guys wore ear plugs even when the machine room doors were closed so that we could concentrate on programming.

Does anyone remember a drum printer that sounded like an M60 machine gun when the door was open, I think it was from DEC.

---

I don't buy computers, I build them!!

Does anyone remember a drum printer that sounded like an M60 machine gun when the door was open, I think it was from DEC.

Phooey. Printing was for the night operator. :)

I do remember the old IBM/360 9-tracks. They sounded like a vacuum cleaner sucking up a Walmart bag when they mounted. Also the 2311 disk drives that would walk around a bit when the seek action got too feisty on a sort/merge.

The good old days...

From Seti@home home page News

News
Server Relocation
We are moving most of our servers to a colocation facility with better cooling, power, and network, which should improve uptime in general. This procedure will start on Monday, April 1st at 08:00 (PDT) and should take 3 days to complete, during which our entire project will be unavailable.
20 Mar 2013, 21:56:25 UTC · Comment

I wonder if they are doing the Boinc Server this week, it seems to have become unavailable, i don't remember a post saying when it was moving,

Claggy

From Seti@home home page News

News
Server Relocation
We are moving most of our servers to a colocation facility with better cooling, power, and network, which should improve uptime in general. This procedure will start on Monday, April 1st at 08:00 (PDT) and should take 3 days to complete, during which our entire project will be unavailable.
20 Mar 2013, 21:56:25 UTC · Comment

I wonder if they are doing the Boinc Server this week, it seems to have become unavailable, i don't remember a post saying when it was moving,

Claggy

They were messing about with the git repo earlier.

Does anyone remember a drum printer that sounded like an M60 machine gun when the door was open, I think it was from DEC.

Phooey. Printing was for the night operator. :)

I do remember the old IBM/360 9-tracks. They sounded like a vacuum cleaner sucking up a Walmart bag when they mounted. Also the 2311 disk drives that would walk around a bit when the seek action got too feisty on a sort/merge.

The good old days...

You will be surprised how much printing was done during the day. Lost several drives because of the walk-about. We upgraded to a 370/45 and replaced the 2311 with the 2314 with the big controllers on the end; I think this was just before we upgraded from DOS to OS/VS, which was the father of MVS. The weight of the double stacked drives stopped the walk-about. Regarding the tapes, I seem to remember that when it was re-winding and the tape got to the end it sounded like a kid with a card stuck in he spokes of the bike going down hill -- hell of a racket.

---

I don't buy computers, I build them!!

How bad is the network congestion right now and how much impact does this mean when it comes to the re-location of the servers?

From what I know, there apparently are some people among the staff that think that they could handle the resources better by moving the servers somewhere else.

There may be several factors in between when it comes to all of this and I have discovered that I have got quite many tasks in recent days, but that it needed several restarts in between, since I now have several discs in my inventory. Also the download of the CUDA tasks are still the bottleneck, even though they are apparently first in the download queue.

The worst thing that can happen is a totally stuck network which can not be blamed on one or a couple of users alone. In the end, you may choose almost every place where business is going on in order to collect or merge a combined network resource, at least when it comes to cities in developed countries.

The next option is to choose what is the best location when it comes to the people who are administering the servers. Are they supposed to climb to the top of hill each morning, or perhaps drive by car or bicycle to the centre of town instead?

The BOINC forums database seems to be overloaded. You can browse around when not logged in, but logging in and then browsing around will take hours. Best not try it at this time.

---

cheers mate ...

You too.

I engage in the discussion because, as a senior member of the crew, I must.
Not everybody agrees with my opinions.

I sometimes have insides to what is happening, but not always.

For example, I was totally left out of the discussion about the relocation.

I asked several questions. But none were answered.

So, I am just another user, as are you.

FWIW as far as I can tell they kept that completely internal.

I still think raising the limits wouldn't achieve much at this point. We are maxxed out and have been for a while. work is going out as fast as the infrastructure supports. If the limits were raised, the high end machines would probably have some more work. at the same time there is only so much cake - your share gets bigger, other shares get smaller. It may distribute over so many machines that it's not very noticeable but stil. Only so much cake.
Whether the throughput would be higher is anyones guess.

Scarecrow used to have that very enlightening graph about what part of the work was done by what RAC groups ;)

I really DO miss Scarecrow's graphs.
I think he lost them due to a crash, and never brought them back.
I asked a few times if he was going to, but never really got an answer.

Been a long time now.

As to the coming outage.......
I guess I am no different than any other user. When it goes down, I got no more work left to process than anybody else.
I must say that I miss the time when I could keep crunching for a freaking fortnight even when the servers went down.
70 or 80 thousand WUs cached. And not a one of them tossed in late.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

How bad is the network congestion right now and how much impact does this mean when it comes to the re-location of the servers?

The current network connection through the Berkeley infrastructure is 100mb, and it's basically running at MAX, 24/7, unless something else is broken. This means that no matter what else they do server wise they can't physically send out more work then they currently are. I'm also guessing that the slow downloads and retries put more load on the download server than if they could send the work down at a decent speed? (more concurrent connections to track etc). And wastes a % of the scarce bandwidth. I'm assuming the co-location site will allow them to utilise the 1gb internet connection that they actually have, and so solve that particular issue.

Other major recent outages have been related to power and air-con, which are probably only "office" standard, and probably a bit old and creaky as well. A modern data centre will have a better standard of these services?

I't wont fix ALL the issues, and some new ones will probably creep out of the woodwork once the network bottleneck is removed. But it should solve 3 of the issues that ARE currently causing problems, and are otherwise out of the projects control (it's not actually their building or network)

Ian

The BOINC forums database seems to be overloaded. You can browse around when not logged in, but logging in and then browsing around will take hours. Best not try it at this time.

Seems to go wonky when it gets to the part of the page where the users PM mail box should display. Overloaded? Corrupted I suspect.

---

That's something new. Earlier I got info back that something unknown was slowing Isaac down, and that they were looking into it. But this loading of the PM box corrupting, perhaps that they're adding the Sent box as well. Dunno. Minecraft is taking up too much of my time. ;-)

---