• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

SetiQ Round Robin Suggestion

J

Jay

Golden Member
As I sit here and look at OrangeKid's que, I see that he's sitting on 1,700+ WU to be uploaded to Berkley. I wonder how much bandiwth and disk space OrangeKid is having to use to run this proxy, not to mention the stress on the machine and the bandwith.

The DNet Round Robin that bphantom helped established really helped TA members when DNet went down, or even if Mika's went down for a short period of time. The ability to share the workload among 3-6 machines vs. one machine definetly was and is illustrated with the round robin we have for the DNet users.

My suggestion is that a similar setup be established for Seti. This way OrangeKid won't be slammed so much with uploading and downloading WU. (I think there is talk of moving the team SetiQ to www.teamanandtech.com once it's up and running at XMission). If that is true, then of course OrangeKid would be a perfect round robin server along with a few others (TACube comes to mind as another possible round robin).

I would go further to say that each Round Robin be given a period of time each day to upload and download to the Main SetiQ server, thus alleviating some of the bandwith being used. This would also help regulate the bandwith going to and from Berkley and maybe help free up some bandwith there as well.

I think it would be easy to establish a SetiQ Status page similar to the one used by the DNet RR as well so that people would be able to monitor the Q's easily. And no, I don't want to be a Round Robin, my server's disk space is too low for that 😉

Jay
 
just a few questions about the SETI queues. Howmuch disk space is needed to run one, how much UL/DL bandwith is recommended, and what are some good system specs (ram,proc) for one?
 
The problem with SETIQueue is that it isn't a "SHARE ALL WU" to anybody....it downloads WU's per user client...so if a person was to bounce around a round-robin, each Queue on the round-robin would have to hold WU's in a sub-Queue for each of those clients.

I agree with the idea of public queues flushing after a certain time....but it would be much better to do it at night because of the Berkeley daytime limits...

SIGH.....

Maybe the weekends will open up....dunno!
 
Jay,
I like your idea, an address that would in turn distribute the load to several machines would be a welcome relief 🙂
These machines would not necessarly have to connect to a single "Team" server, I would think that Berkeley would do well as a main connecting point.
By doing this it would greatly ensure that there is plenty of work available for all.
Currently with the situation at Berkeley some people are 'hoarding' work in LARGE quantities and depriving some of our smaller users of the ability to get work.....I have watched 500~1000 wus go out to people (Q's) that have maybe three clients using it.........seems a bit outlandish.
Most people get their "stats fix" directly from Berkeley so a loss of some personal stats would be of small consiquence for the greater good of all.
I'm sure people could use a paticular Q all the time if they really wished, but again I have to agree that the time has come to spread out the load.

SetiQ seems to require intervention from time to time when dealing with large amounts of users........this, for those that would like to participate in something of this nature.

I will continue to run the Q at my address either way

Just call me the BlueBird of happiness
 
Jay - I remember a discussion about that a long time ago, but then it sortof peetered out... 🙁

BFC - Each SETI WU = ~360K of space. So as a minimum, if you multiply the number of expected WUs times that size, you'll start to get a little idea. Plus you'll need to take into account the log files & results files (that will both vary in size but aren't big), client ini files, the Setiqueue config files and binary, a "dropped" folder with files, etc.

My Setiqueue currently has ~265 WUs in it and I have it set to save my results. This queue (me having switched to the new one from the old one) has been running since ~March 2001. The directory is currently ~105MB in size.

As for bandwidth, you might want to ask OK what he has on his connection. Right now, I don't think the bandwidth to the queues is the issue, it's the bandwidth (or better, lack of sockets/connections) getting stuff from SETI... 🙁
 
Currently my Q runs between 3 and 5 GB depending on the amount of work queued

My connection is a cable connection with around 300kbps upstream and around 1000kbps down

The box i'm using is dual p3-850's with 512 meg ram and a 40 gig hd (10/30)......OS on one partition and public stuff on the other.........it is also running a pproxy for dnet clients, is my web server and NAT box..... and does rc5 on the side 🙂
 
Just another suggestion. What if we get that list out of all the available Queues and try and distribute users to them.

I have been using OK's Queue for quite awhile until the Orange crush 🙁 I went back to OK's Queue until the bandwidth problem.

Now that I am getting units from Berkley directly I am starting to send in Duplicate results🙁

The Queues are Great for the Team. Let use them wisely.🙂

-Tarca
 
Just another suggestion. What if we get that list out of all the available Queues and try and distribute users to them.

The queue list is in a thread being bumped right now! 😀
 
Engineer said:
------------------------


<< The problem with SETIQueue is that it isn't a "SHARE ALL WU" to anybody....it downloads WU's per user client...so if a person was to bounce around a round-robin, each Queue on the round-robin would have to hold WU's in a sub-Queue for each of those clients. >>


------------------------

Sounds like a round robin would not really work effeciently with SETI. With this in mind, what if we distributed the user between the various queues rather than the huge load we have on OK's Proxy? Each member could be assign, or invited to join a specific proxy. If one starts to get overloaded the excess could be redistributed.

/me thinks I was a bit slow with this post...was watching TV and trying to compose
 
JWM - I think at this point, that's about the only thing we can do... 🙁

I was even going to suggest that people temporarily pull back completely from OK's for a day or so until he's flushed most of the results and then slowly come back or go to other queues in a more controlled fashion. 🙂
 
^
|
|
|
|
rolleye.gif
 
It would be cool to have some kind of system where clients would be set to something like proxy.teamanandtech.com:5001, and there would be some kind of list of users, and users would be assigned to different queues. Therefore when a person dumps to proxy.teamanandtech.com:5001 they would be routed to the queue they're assigned to. I don't know if anything like that would work, just a thought 🙂.
 
hmm yeah that would be cool. It would also be cool if we could have some sort of default backup que. I know if you use dns2go then it has an option for offline forwarding, so it would be cool if there was a way of forwarding people to a reserve que should our friend OK go down or be overloaded. hmm just ideas
rolleye.gif
(sorry just saw this new emotion and wanted to try it out 😉)
 
OK, would you prefer it if I pointed my personal queue directly to berkeley for now?I like the double stat's fix, but I dont want to cause you any problem... what do you think?
 
I think that having the list of TeAm Queues available like it is will help. I don't see a more complex routing system working, unfortunately, as it may just cause confusion. 😕

The largest problem I've seen lately has been new users trying to connect to a SetiQ, but Berkeley not accepting the passthrough. It's been a huge pain for a lot of people, as they don't know that with all Berkeley's recent troubles, it take 4-5 tries to get through.

rolleye.gif
<----- hehe, cool!
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top