Vent away,,,,get it off your chest... Since I started collecting data, every several days I end up setting my hosts to No New Work, so I can drain them before switching apps, then when I try to get more work....... I feel your pain. actually, last week it's seems to work OK. We have noticed that when these quick wus are split from the discs, that everyone is normally requesting more work units to fill the same period of time, even though they have limited storage for "wus ready to send", so When they send out "shorties", they always run dry (or atleast they have the last few times). The ONLY way to assure work is to set a higher cache. Higher caches are generally frowned upon by some users, so try to find something that works for YOU. If two doesn't work across all AR's then try 3, If that doesn't work, go for 4, then 5 etc, until you can weather any issue.

Currently the servers are running, it's just that we're sucking the work up faster than they can stock it up.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I do infact know. I just wrote about it a minute ago in this post to another user.

I imagine if they had enough hard drive space say 50 Terabytes, and enough servers to "presplit and store" wus, then they might be able to sort the work by AR and distribute according to your idea.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

Why should they? The "20-minute" short workunits need processing just as much as the long ones do. The Project hasn't guaranteed that there will be work available for any particular participant to download at any given time. In fact, they have stated that there will be times when there won't be.


There is data being split from the tapes. The (currently) 5 MultiBeam splitters are running, grinding out over 12 results per second. Someone is getting these. They are just going out as fast as they are being split, so there is no surplus being built up. There are times (like now) that, due to the data being split (short-run Angle Ranges and noisy (-9) workunits), the total processing capacity of the computers attached to S@H exceeds the amount of data that the project can supply. While the project staff are working to increase their splitting capacity (too many splitters running currently bog down their internal networks and servers), for now we just have to deal with it. But even that is going to eventually have its limits, and with ever-increasing processing capacity among the participants, we are going to hit the wall once again... on a permanent basis... sooner or later...

This is why EVERY participant in S@H/BOINC is strongly encouraged to attach to another BOINC project (or projects) in addition to S@H -- to keep their computers from going idle when S@H is 'out of work' and share out the processing capacity to other, equally deserving projects. Those who choose to not do so (and run S@H only) have effectively consented to having their computers be idle when S@H is 'out of work'.

---

https://youtu.be/iY57ErBkFFE

#Texit

Don't blame me, I voted for Johnson(L) in 2016.

Truth is dangerous... especially when it challenges those in power.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

My fastest box just spent the last 2 days crunching nothing except the short 20 minute WU's in its cache. Finished a couple of hours ago. Now my next fastest box started doing the same thing and I expect it to take 2 days for it to complete all of it's 20 minute WU's. Not complaining as it is part of carrying a cache on the computers and I pride myself in crunching all the work assigned to me and my boxes.

---

Boinc....Boinc....Boinc....Boinc....

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

My fastest box just spent the last 2 days crunching nothing except the short 20 minute WU's in its cache. Finished a couple of hours ago. Now my next fastest box started doing the same thing and I expect it to take 2 days for it to complete all of it's 20 minute WU's. Not complaining as it is part of carrying a cache on the computers and I pride myself in crunching all the work assigned to me and my boxes.

Well that's my problem, I'm crunching it all and then some LOL. I don't mind the small WUs as long as my machines can get 'em. But everyone tears through them so fast these days.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

RAC upto 1620ish now and pending upto 7500+

[edit] oh and three machines bone dry again [/edit]

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

Great thought, but I'm afraid they probably need more than just adapters. Have a look at this thread - it started life here, but got moved to the Staff Blog for some reason. The servers they need to replace are:

# kosh: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# klaatu: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# penguin: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)

- could you check whether your carton has Solaris drivers on the CD? <grin>

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

Great thought, but I'm afraid they probably need more than just adapters. Have a look at this thread - it started life here, but got moved to the Staff Blog for some reason. The servers they need to replace are:

# kosh: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# klaatu: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# penguin: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)

- could you check whether your carton has Solaris drivers on the CD? <grin>

Hrmmm... well I have an HP Proliant DL 145 1U server sitting on my desk doing diddly squat. If they could arrange to cover the shipping it would be theres. nice box. I think I put 2 1.6 gig Opterons in it (it's too loud to run as a cruncher LOL). 2 or 4 gigs of ram, but ram is cheap.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

Great thought, but I'm afraid they probably need more than just adapters. Have a look at this thread - it started life here, but got moved to the Staff Blog for some reason. The servers they need to replace are:

# kosh: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# klaatu: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# penguin: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)

- could you check whether your carton has Solaris drivers on the CD? <grin>

Hrmmm... well I have an HP Proliant DL 145 1U server sitting on my desk doing diddly squat. If they could arrange to cover the shipping it would be theres. nice box. I think I put 2 1.6 gig Opterons in it (it's too loud to run as a cruncher LOL). 2 or 4 gigs of ram, but ram is cheap.

Well, shipping from Canada probably makes more sense than shipping from the UK. If you're serious, why not drop Blurf a private message? He might be able to pair you up with another donor who could pick up the tab for the shipping, and a third to toss in a bit of extra RAM.

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

Great thought, but I'm afraid they probably need more than just adapters. Have a look at this thread - it started life here, but got moved to the Staff Blog for some reason. The servers they need to replace are:

# kosh: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# klaatu: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# penguin: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)

- could you check whether your carton has Solaris drivers on the CD? <grin>

Hrmmm... well I have an HP Proliant DL 145 1U server sitting on my desk doing diddly squat. If they could arrange to cover the shipping it would be theres. nice box. I think I put 2 1.6 gig Opterons in it (it's too loud to run as a cruncher LOL). 2 or 4 gigs of ram, but ram is cheap.

Well, shipping from Canada probably makes more sense than shipping from the UK. If you're serious, why not drop Blurf a private message? He might be able to pair you up with another donor who could pick up the tab for the shipping, and a third to toss in a bit of extra RAM.

Already did. and frankly I'd like to have it off my desk. If someone covers the shipping I might even be able to toss in another couple gigs of ram. It'll bury those old Sparc machines, that much I know.

[edit: 5 machines now offline due to no work units]

Now I know you probably don't know, but why the heck are the splitters turning out the little short 20 minute WUs. There has to be some way SETI can turn out the WUs based on what's in storage. If the queue gets short, turn out the long ones, if the queue is almost full top it up with little ones. But keep track of the top ups so you don't end up with a full queue of little guys.

I had much the same idea here, and asked Matt if he knew in advance what sort of work was included in each recording session. I don't think he's replied.

As others have commented, it's not a very good long-term solution, but keeping a balance of long and short work going through the splitters would help to keep the stress levels down while the hardware donations drive (they need new servers with Gigabit ethernet) is gathering strength.

If they need gigabit ehternet adapaters I'll send them a box full. Seriously, if that will solve their problems I'll ship em a crate of the damned things (or at least a carton).

Great thought, but I'm afraid they probably need more than just adapters. Have a look at this thread - it started life here, but got moved to the Staff Blog for some reason. The servers they need to replace are:

# kosh: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# klaatu: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)
# penguin: Sun D220R (2 x 440MHz Sparc, 2 GB RAM)

- could you check whether your carton has Solaris drivers on the CD? <grin>

Hrmmm... well I have an HP Proliant DL 145 1U server sitting on my desk doing diddly squat. If they could arrange to cover the shipping it would be theres. nice box. I think I put 2 1.6 gig Opterons in it (it's too loud to run as a cruncher LOL). 2 or 4 gigs of ram, but ram is cheap.

Well, shipping from Canada probably makes more sense than shipping from the UK. If you're serious, why not drop Blurf a private message? He might be able to pair you up with another donor who could pick up the tab for the shipping, and a third to toss in a bit of extra RAM.

Already did. and frankly I'd like to have it off my desk. If someone covers the shipping I might even be able to toss in another couple gigs of ram. It'll bury those old Sparc machines, that much I know.

Great stuff. One down, two to go. Anybody else got a spare server cluttering up their workbench?

Great stuff. One down, two to go. Anybody else got a spare server cluttering up their workbench?

Well I am really surprised it's that hard for them to find hardware. If I get a response to my private message, I'll see if there are other ways I can help as well. I know some folks at some various manufacturers, I'm sure they can spare some hardware for the cause. (good corporate citizenship and what not).

On the plus side my RAC is upto 1728 and my pending is holding steady around 7500 units.

On the down side I still have 4 machines out of work :/

Well I am really surprised it's that hard for them to find hardware.

Getting hardware is part of the problem, the other problem is getting hardware that is compatible with all the other hardware received.

---

Grant
Darwin NT

Well I am really surprised it's that hard for them to find hardware.

Getting hardware is part of the problem, the other problem is getting hardware that is compatible with all the other hardware received.

I just have a hard time believing that one of the big guns out there would have an issue being a corporate sponsor/donor for SETI@Home. Intel, AMD, Microsoft, etc. all have pockets so deep that they would never miss the few thousand dollars it would cost them to support SETI. And the publicity in front of SETI users would certainly be worth the investment.

Well I am really surprised it's that hard for them to find hardware.

Getting hardware is part of the problem, the other problem is getting hardware that is compatible with all the other hardware received.

I just have a hard time believing that one of the big guns out there would have an issue being a corporate sponsor/donor for SETI@Home. Intel, AMD, Microsoft, etc. all have pockets so deep that they would never miss the few thousand dollars it would cost them to support SETI. And the publicity in front of SETI users would certainly be worth the investment.

Part of the problem is that there's a lot of people out there that feel the Search for Extraterrestrial Intelligence is a waste of time at this point (no offense to SETI, I'm a proud member myself obviously), especially considering the extreme probability that we will not find anything given our current search methods, many feel that solving our own problems such as with Folding@Home and CPDN are much more worthwhile tasks (considering everyone is concerned about going green these days). Donating servers to SETI@Home may not offer as much face time as you'd think, given the current social and political climate.

---

Another huge problem is, that there doesn't exits a welldocumented flowchart of the serverpark.

I don't think that there exits any documentation at all!
Many people have asked for this over the years, as to be able to offer their help and place accurate hardware-donation.

If one or two key-serverstaff-members leave the project for good (in anger og accidential) the whole projects shuts down IMHO.

My 2 cent.

Kiva

---

Greetings from Norway

Crunch3er & AK-V8 Inside

Osiris30

As a committed seti cruncher, well done on the donation.

---

Another huge problem is, that there doesn't exits a welldocumented flowchart of the serverpark.

I don't think that there exits any documentation at all!
Many people have asked for this over the years, as to be able to offer their help and place accurate hardware-donation.

If one or two key-serverstaff-members leave the project for good (in anger og accidential) the whole projects shuts down IMHO.

My 2 cent.

Kiva

Probably the only really key staff member is Eric Korpela, being the project scientist and all. There are at least three other staff members that know as well, we just don't know about them because they don't post. Matt mentions them from time to time.

The BOINC structure sets out the logical servers (upload servers, download servers, schedulers) pretty well. The rest can be found by reading configuration files and DNS and mounts.

---