Successful Lab Repairs


log in

Advanced search

Message boards : News : Successful Lab Repairs

Previous · 1 · 2 · 3 · Next
Author Message
Profile Gary Charpentier
Volunteer tester
Avatar
Send message
Joined: 25 Dec 00
Posts: 11732
Credit: 5,969,877
RAC: 0
United States
Message 1341706 - Posted: 28 Feb 2013, 18:29:54 UTC

There is a strong perception there is a bandwidth problem by those who want high RAC's. I've asked several times but the lab boys don't answer is are they collecting data at a much faster rate than we are processing it. If not, then in reality there isn't a bandwidth issue.

____________

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 37392
Credit: 500,213,525
RAC: 526,140
United States
Message 1341709 - Posted: 28 Feb 2013, 18:43:01 UTC - in response to Message 1341706.
Last modified: 28 Feb 2013, 18:44:45 UTC

There is a strong perception there is a bandwidth problem by those who want high RAC's. I've asked several times but the lab boys don't answer is are they collecting data at a much faster rate than we are processing it. If not, then in reality there isn't a bandwidth issue.

And there is a strong perception by some with low RACs that there is simply no bandwidth problem at all. Or that those with high RACs are more interested in their stats than the science. Both are untrue, at least speaking for myself.

However, your question is a valid one, and I would be interested in the answer as well.
____________
******************
Seti whacko, resident evil, and town clown...

Crunching Seti, loving all of God's kitties.

I have met a few friends in my life.
Most were cats.

juan BFB
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 4619
Credit: 233,720,203
RAC: 347,967
Brazil
Message 1341711 - Posted: 28 Feb 2013, 18:52:04 UTC - in response to Message 1341700.
Last modified: 28 Feb 2013, 19:02:28 UTC

Interesting, it never seemed to me that there was a bandwidth problem. To me donwloading SAH takes seconds, AP minutes.

You win the lottery! We all have problems to DL WU from SETI.

There is a strong perception there is a bandwidth problem by those who want high RAC's.

Of course if you have a high RAC you number of DL WU needed to keep your hosts is bigger, but i realy don´t understand why now a high RAC is Bad?.

I agree 100% with Mark, if RAC is what we are looking for, there are a lot of others projects who "paid" more RAC for the same amount of processing power and electric power we spend to mantain our RAC, i could say only for me, i just here because a dream, and thats all, nothing about science, RAC or anything else just about a dream to answer the old question: are we alone?

On other hand, someone loose the focus, yes there is a bandwidth problem, anyone could see by the crickets the link is uses at 100% most of the time, and anybody who knows a little about network usage knows, no network works fine when you use 100% of it´s capacity.
____________

Profile Gary Charpentier
Volunteer tester
Avatar
Send message
Joined: 25 Dec 00
Posts: 11732
Credit: 5,969,877
RAC: 0
United States
Message 1341801 - Posted: 28 Feb 2013, 23:43:31 UTC - in response to Message 1341709.

There is a strong perception there is a bandwidth problem by those who want high RAC's. I've asked several times but the lab boys don't answer is are they collecting data at a much faster rate than we are processing it. If not, then in reality there isn't a bandwidth issue.

And there is a strong perception by some with low RACs that there is simply no bandwidth problem at all. Or that those with high RACs are more interested in their stats than the science. Both are untrue, at least speaking for myself.

However, your question is a valid one, and I would be interested in the answer as well.

No one is saying that people with high RAC's aren't interested in the science. However there are those that seem as if they are chasing RAC more than they are interested in science, or that they know better than the project scientist what data should be sent out and how fast.

As to the question being valid, if we are crunching faster than it is being collected, then eventually the well runs dry and bandwidth is moot.

In any case the possible move to a co-lo spot on campus should remove one of the two political reasons for the bandwidth limit. Unfortunately the other political reason is $$$/month based and may not be so easy to overcome, unless someone can get it buried deep in a budget.

____________

Rolf
Send message
Joined: 16 Jun 09
Posts: 114
Credit: 7,800,622
RAC: 2
Switzerland
Message 1341811 - Posted: 1 Mar 2013, 0:14:33 UTC - in response to Message 1341801.

unless someone can get it buried deep in a budget.


This way, it will stay political for eternity!

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 37392
Credit: 500,213,525
RAC: 526,140
United States
Message 1341879 - Posted: 1 Mar 2013, 7:15:33 UTC - in response to Message 1341801.

There is a strong perception there is a bandwidth problem by those who want high RAC's. I've asked several times but the lab boys don't answer is are they collecting data at a much faster rate than we are processing it. If not, then in reality there isn't a bandwidth issue.

And there is a strong perception by some with low RACs that there is simply no bandwidth problem at all. Or that those with high RACs are more interested in their stats than the science. Both are untrue, at least speaking for myself.

However, your question is a valid one, and I would be interested in the answer as well.

No one is saying that people with high RAC's aren't interested in the science. However there are those that seem as if they are chasing RAC more than they are interested in science, or that they know better than the project scientist what data should be sent out and how fast.

As to the question being valid, if we are crunching faster than it is being collected, then eventually the well runs dry and bandwidth is moot.

In any case the possible move to a co-lo spot on campus should remove one of the two political reasons for the bandwidth limit. Unfortunately the other political reason is $$$/month based and may not be so easy to overcome, unless someone can get it buried deep in a budget.

Well, my opinion is let's get the infrastructure capable of handling as much as the project can dish out. If that means that we are processing more data than the project is gathering...I think that would be a wonderful thing.
Then room would be open to either gather more data if possible, or process the data in more depth than is currently being done.
Either way....the bandwidth we now have is not being used to it's fullest, as most know that when maxxed out like this, the end result is less data actually being transferred than it appears, as much is lost to retries, transmission errors, etc..
If we had a 200Mb pipe, I think it would solve a lot of problems. More would be better, of course, but I would not be surprised if maximum usage ended up to be 200Mb or less after an outage, and settling in at not much over the 100Mb we have now, but it would be fully utilized, not wasted with corrupted data transfer attempts.

____________
******************
Seti whacko, resident evil, and town clown...

Crunching Seti, loving all of God's kitties.

I have met a few friends in my life.
Most were cats.

gs
Volunteer tester
Avatar
Send message
Joined: 5 Jan 07
Posts: 29
Credit: 2,661,475
RAC: 2,174
Germany
Message 1342342 - Posted: 2 Mar 2013, 9:36:55 UTC

It seems that the capacity of the servers still is degraded. I noticed, that during the last weeks, DL of astropulse and standard 6.03 work units takes hours.
I'm DL from Europe, but even during the early morning hours local time, when the activities on the US side should be low, DL will not speed up.
I'm constantly running out of work units.

Is this a persistant problem?

msattler
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 37392
Credit: 500,213,525
RAC: 526,140
United States
Message 1342344 - Posted: 2 Mar 2013, 9:49:17 UTC - in response to Message 1342342.

It seems that the capacity of the servers still is degraded. I noticed, that during the last weeks, DL of astropulse and standard 6.03 work units takes hours.
I'm DL from Europe, but even during the early morning hours local time, when the activities on the US side should be low, DL will not speed up.
I'm constantly running out of work units.

Is this a persistant problem?

You just noticed this...now? LOL.
Yes, it is a persistent problem.
We have been debating about it for years now.
____________
******************
Seti whacko, resident evil, and town clown...

Crunching Seti, loving all of God's kitties.

I have met a few friends in my life.
Most were cats.

gs
Volunteer tester
Avatar
Send message
Joined: 5 Jan 07
Posts: 29
Credit: 2,661,475
RAC: 2,174
Germany
Message 1342368 - Posted: 2 Mar 2013, 14:05:55 UTC - in response to Message 1342344.

It wasn't that much of a problem up to some months ago.
It's getting worse and my question is, what will be done about it?
If bandwidth is not the bottleneck, is it the server capacity?

Rolf
Send message
Joined: 16 Jun 09
Posts: 114
Credit: 7,800,622
RAC: 2
Switzerland
Message 1342375 - Posted: 2 Mar 2013, 14:54:14 UTC - in response to Message 1342368.

It wasn't that much of a problem up to some months ago.
It's getting worse and my question is, what will be done about it?
If bandwidth is not the bottleneck, is it the server capacity?

+1

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5566
Credit: 51,427,216
RAC: 42,192
Australia
Message 1342588 - Posted: 3 Mar 2013, 2:24:19 UTC - in response to Message 1342368.

It wasn't that much of a problem up to some months ago.

It's been a problem for almost 2 years.

____________
Grant
Darwin NT.

Oldskool J-EDM
Send message
Joined: 23 Jan 13
Posts: 3
Credit: 94,020
RAC: 0
United Kingdom
Message 1343023 - Posted: 4 Mar 2013, 18:48:57 UTC

In response to Garys post on this page, I am also having the exact same problem.
I am also based here in the UK.

I'm a noob, joined only a month or so ago...so I do not know much about the day-to-day runnings of SETI just yet.

However...my first week using Boinc (exclusively for SETI) gave me many, many tasks...all DL'd very quickly indeed. Now I seem to get appox 50% of a task before it grinds to a halt...and then, as others have said, can take on average 6-8 hours for the rest to DL'd but I am also only getting 2, maybe 3 tasks at a time...whereas I had initially been recieving a minimum of 8 each time boinc connected to the servers....so my task list kept a nice stockpile to be getting on with.

I'm not a techy, but I do find this both curios and frustrating.

Did I join SETI on a 'good week' then?...or is this a quirk for newcomers to the project?

Profile Gary Charpentier
Volunteer tester
Avatar
Send message
Joined: 25 Dec 00
Posts: 11732
Credit: 5,969,877
RAC: 0
United States
Message 1343079 - Posted: 4 Mar 2013, 21:26:02 UTC - in response to Message 1343023.

In response to Garys post on this page, I am also having the exact same problem.
I am also based here in the UK.

I'm a noob, joined only a month or so ago...so I do not know much about the day-to-day runnings of SETI just yet.

However...my first week using Boinc (exclusively for SETI) gave me many, many tasks...all DL'd very quickly indeed. Now I seem to get appox 50% of a task before it grinds to a halt...and then, as others have said, can take on average 6-8 hours for the rest to DL'd but I am also only getting 2, maybe 3 tasks at a time...whereas I had initially been recieving a minimum of 8 each time boinc connected to the servers....so my task list kept a nice stockpile to be getting on with.

I'm not a techy, but I do find this both curios and frustrating.

Did I join SETI on a 'good week' then?...or is this a quirk for newcomers to the project?


Interesting ... this morning DL'd 70 work units PDQ. There were a couple that timed out, but by the time the rest had gone, they went right away without intervention.

I'm beginning to wonder if what people are seeing is a related to a wonky TCP/IP stack or overloaded routers somewhere between the lab and them. I'm wondering if the timeouts are due to excessive latency in the route causing one end or the other to timeout prematurely.

I'm thinking this because if it is the same people having trouble, while others have no trouble then it may be related to the path or time the packets are taking to the lab. I know we are slamming the lab, but you would expect that to randomly cause issues, not for the same people most of the time.

____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5566
Credit: 51,427,216
RAC: 42,192
Australia
Message 1343189 - Posted: 5 Mar 2013, 6:24:20 UTC - in response to Message 1343079.


Add to that fact that those having issues, if they use a proxy, will then get excelent download rates, till the proxy gets closed down.
____________
Grant
Darwin NT.

mbondaz
Send message
Joined: 28 Oct 12
Posts: 1
Credit: 183,857
RAC: 336
France
Message 1343329 - Posted: 5 Mar 2013, 23:40:16 UTC - in response to Message 1343189.

The problem appears ans disapears without any changes of network triming.

For example, it seems a quiet slow :

SET@home|setiathome-5.28x86_64-pc-linux-gnu|19,67%|1,25/6,33 MB|38:15:24|0,00KBps|Téléchargement: réessayer dans 05:21:50

Sean Glasspool
Avatar
Send message
Joined: 23 May 99
Posts: 7
Credit: 6,984,179
RAC: 1,680
New Zealand
Message 1343879 - Posted: 7 Mar 2013, 23:01:20 UTC - in response to Message 1343329.

Im having the same issue here. WUs sitting mostly downloaded then stalled. I connect to a proxy service and they start downloading again, but very slowly...
____________

juan BFB
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 4619
Credit: 233,720,203
RAC: 347,967
Brazil
Message 1343881 - Posted: 7 Mar 2013, 23:05:25 UTC
Last modified: 7 Mar 2013, 23:45:54 UTC

If you have Windows hosts and have DL problems, you shoulD read this thread:

http://setiathome.berkeley.edu/forum_thread.php?id=71002
____________

bill
Send message
Joined: 16 Jun 99
Posts: 848
Credit: 20,600,241
RAC: 14,757
United States
Message 1343882 - Posted: 7 Mar 2013, 23:09:42 UTC - in response to Message 1343879.
Last modified: 7 Mar 2013, 23:10:05 UTC

There's an answer to the stall problem being passed out in
Number Crunching for the Windows users. Funny that, getting a fix
to a problem in a forum for problems instead of one for news.

Richard Haselgrove
Volunteer tester
Send message
Joined: 4 Jul 99
Posts: 8275
Credit: 44,971,001
RAC: 13,888
United Kingdom
Message 1343904 - Posted: 8 Mar 2013, 0:06:47 UTC - in response to Message 1343882.

There's an answer to the stall problem being passed out in
Number Crunching for the Windows users. Funny that, getting a fix
to a problem in a forum for problems instead of one for news.

Yes, it's nice when volunteers are part of the solution, not just part of the problem.

I'm a great believer in the principle of BOINC being a provider of Distributed Wisdom, not just a consumer of Distributed Computing.

bill
Send message
Joined: 16 Jun 99
Posts: 848
Credit: 20,600,241
RAC: 14,757
United States
Message 1343919 - Posted: 8 Mar 2013, 1:28:13 UTC - in response to Message 1343904.

Can we get a Hallelujah?

Previous · 1 · 2 · 3 · Next

Message boards : News : Successful Lab Repairs

Copyright © 2014 University of California