Message boards : Number crunching : dear Neasan & Alex
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile Neasan
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 30 Nov 06
Posts: 234
Credit: 11,078
RAC: 0
Message 17917 - Posted: 18 Sep 2007, 13:54:48 UTC - in response to Message 17916.  

You are correct about the luck element this is why we tried to drop quotas and up the amount of WUs so they would be on the server long enough for either people to see them and poll themselves or for their managers to do so.
Or I played some games with the BOINC manager and forced it out of cycle.

I assume people did just this when they saw some WUs would be released.
ID: 17917 · Report as offensive     Reply Quote
Profile [B^S] Dora

Send message
Joined: 27 Apr 06
Posts: 26
Credit: 13,559
RAC: 0
Message 17918 - Posted: 18 Sep 2007, 14:57:13 UTC

Can you give us a quota and then block us for 10 minutes or some other specified period of time before we can poll again?

I found an old SETI post that indicates that this may be possible...

I think they did this in answer to huge number of polls after they came back online after maintenance:


4/25/06 8:12:33 AM|SETI@home|Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
4/25/06 8:12:33 AM|SETI@home|Reason: To fetch work
4/25/06 8:12:33 AM|SETI@home|Requesting 209942 seconds of new work
4/25/06 8:12:39 AM|SETI@home|Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
4/25/06 8:12:39 AM|SETI@home|Message from server: Not sending work - last request too recent: 233 sec
4/25/06 8:12:39 AM|SETI@home|No work from project
4/25/06 8:19:51 AM|SETI@home|Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
4/25/06 8:19:51 AM|SETI@home|Reason: To fetch work
4/25/06 8:19:51 AM|SETI@home|Requesting 209941 seconds of new work
4/25/06 8:19:57 AM|SETI@home|Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
4/25/06 8:19:57 AM|SETI@home|Message from server: Not sending work - last request too recent: 439 sec
4/25/06 8:19:57 AM|SETI@home|No work from project
4/25/06 8:20:07 AM|SETI@home|Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
4/25/06 8:20:07 AM|SETI@home|Reason: To fetch work
4/25/06 8:20:07 AM|SETI@home|Requesting 209940 seconds of new work
4/25/06 8:20:12 AM|SETI@home|Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
4/25/06 8:20:12 AM|SETI@home|Message from server: Not sending work - last request too recent: 15 sec
4/25/06 8:20:12 AM|SETI@home|No work from project
4/25/06 8:30:23 AM|SETI@home|Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
4/25/06 8:30:23 AM|SETI@home|Reason: To fetch work
4/25/06 8:30:23 AM|SETI@home|Requesting 210250 seconds of new work
4/25/06 8:30:28 AM|SETI@home|Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
4/25/06 8:30:30 AM|SETI@home|Started download of 04oc99aa.22368.881.748586.1.143
4/25/06 8:30:30 AM|SETI@home|Started download of 16au99ab.19802.11282.754822.1.62
4/25/06 8:32:34 AM|SETI@home|Finished download of 04oc99aa.22368.881.748586.1.143
4/25/06 8:32:34 AM|SETI@home|Throughput 2942 bytes/sec
4/25/06 8:32:36 AM||request_reschedule_cpus: files downloaded
4/25/06 8:32:53 AM|SETI@home|Finished download of 16au99ab.19802.11282.754822.1.62
4/25/06 8:32:53 AM|SETI@home|Throughput 2552 bytes/sec
4/25/06 8:32:55 AM||request_reschedule_cpus: files downloaded

The reply was from a volunteer tester:

There is a 10 minute 4 second back off, to prevent users from continually hitting the update button. you have to wait or it will stop you. It's there to lessen the load on the servers.




ID: 17918 · Report as offensive     Reply Quote
PovAddict
Avatar

Send message
Joined: 14 Jul 05
Posts: 275
Credit: 49,291
RAC: 0
Message 17919 - Posted: 18 Sep 2007, 16:32:48 UTC - in response to Message 17915.  
Last modified: 18 Sep 2007, 16:36:36 UTC

There are two quotas:
1) The amount of WUs you can download at a time
2) The number of WUs you can download for each CPU


I'll clarify that... There are four parameters that can be tweaked:
1a. The maximum amount of WUs you can download on a single scheduler request. (max_wus_to_send)
1b. The minimum amount of time you must wait between requests. (min_sendwork_interval)
2. The maximum amount of WUs you can download for each computer *per day*. The counter to check for this quota is reset (for that host) every 24 hours. (daily_result_quota)
3. The maximum amount of WUs a computer can have on its cache at a time (max_wus_in_progress; not supported by your ANCIENT server code).

A high min_sendwork_interval and low max_wus_to_send would improve things quite a bit. But the real solution is max_wus_in_progress. That would really avoid hosts with a hundred vs hosts with nothing.
ID: 17919 · Report as offensive     Reply Quote
PovAddict
Avatar

Send message
Joined: 14 Jul 05
Posts: 275
Credit: 49,291
RAC: 0
Message 17923 - Posted: 19 Sep 2007, 4:51:32 UTC - in response to Message 17922.  

It would fix another problem too. It would eliminate the need for an initial replication of 5 when a quorum of only 3 is required which would eliminate the ridiculous waste of CPU cycles.

Which means more CPU cycles for other projects. Good point.

How difficult is it to tweak the parameters? Is it just a matter of opening a config file in a text editor and altering a few numbers?

Yes, it is. Just change the number inside the tags <max_wus_to_send>20<max_wus_to_send>.

For max_wus_in_progress, it's a totally different story, since they need to upgrade the server code. Means merging their custom changes with code that has had two years of other changes, for example.
ID: 17923 · Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 24 Nov 06
Posts: 76
Credit: 7,914,481
RAC: 27,114
Message 17924 - Posted: 19 Sep 2007, 5:39:39 UTC

Server SW upgrade would also allow "server side abortion", which would further reduce wasted CPU cycles.
Dublin, California
Team: SETI.USA

ID: 17924 · Report as offensive     Reply Quote
PovAddict
Avatar

Send message
Joined: 14 Jul 05
Posts: 275
Credit: 49,291
RAC: 0
Message 17925 - Posted: 19 Sep 2007, 6:04:23 UTC - in response to Message 17924.  
Last modified: 19 Sep 2007, 6:09:55 UTC

Server SW upgrade would also allow "server side abortion", which would further reduce wasted CPU cycles.

And that's only one of the many. Here are some random ones (I guarantee the list is totally incomplete):

- Server-side workunit aborting
- max_wus_in_progress
- Private messages in forum
- Team "takeover" request
- Working forum rate buttons
- Better forum moderating abilities
- Link to your credits on other projects from your account page and user pages
- New global settings (users currently have to use another project to set all the prefs that LHC doesn't show)
- Performance optimizations on server, particularly on database queries
- Many security fixes on server code (I know details on many of the vulnerabilities, and some are quite bad. But don't even think I'll "teach" you how to exploit them. Eat some bricks instead.)

EDIT: some fairly recent ones, from ServerUpdates:
- "Merge hosts by name" function
- Let team founders view history of people joining/quitting team
- "Mark all threads as read" feature on forums
ID: 17925 · Report as offensive     Reply Quote
Christoph

Send message
Joined: 25 Aug 05
Posts: 69
Credit: 306,627
RAC: 0
Message 17930 - Posted: 19 Sep 2007, 20:12:17 UTC
Last modified: 19 Sep 2007, 20:14:17 UTC

What is the schedule for updating the Server Software?? Please give us some hope. (I'm waiting to merge my hosts)

Happy crunching, Christoph
Christoph
ID: 17930 · Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 6 Jul 06
Posts: 108
Credit: 661,871
RAC: 196
Message 17940 - Posted: 20 Sep 2007, 12:20:29 UTC - in response to Message 17930.  

What is the schedule for updating the Server Software?? Please give us some hope. (I'm waiting to merge my hosts)

Happy crunching, Christoph


Yes agreed.
With Einstein, Cosmology and QMC on the latest Server Version of 601 (also Seti).
Superlink on Server Version 511.
And Rosetta, Ralph, The Lattice Project and CPDN all on 509.
You get poor old LHC bringing up the rear on Server Version 505.

Definately time for an update, even if it just for the new security features.
ID: 17940 · Report as offensive     Reply Quote
PovAddict
Avatar

Send message
Joined: 14 Jul 05
Posts: 275
Credit: 49,291
RAC: 0
Message 17941 - Posted: 20 Sep 2007, 14:11:02 UTC - in response to Message 17940.  

With Einstein, Cosmology and QMC on the latest Server Version of 601 (also Seti).
Superlink on Server Version 511.
And Rosetta, Ralph, The Lattice Project and CPDN all on 509.
You get poor old LHC bringing up the rear on Server Version 505.


That version number doesn't really tell you much. It's actually the client version number that was available along with the server code... Have a look at the revision number on server status page instead. It's only on newer versions, so it's not on LHC, and some projects with customized status pages don't have it (like SETI). However, by looking somewhere else I can tell you SETI has revision 13586 or higher.
ID: 17941 · Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 6 Jul 06
Posts: 108
Credit: 661,871
RAC: 196
Message 17948 - Posted: 20 Sep 2007, 23:05:20 UTC - in response to Message 17941.  

With Einstein, Cosmology and QMC on the latest Server Version of 601 (also Seti).
Superlink on Server Version 511.
And Rosetta, Ralph, The Lattice Project and CPDN all on 509.
You get poor old LHC bringing up the rear on Server Version 505.


That version number doesn't really tell you much. It's actually the client version number that was available along with the server code... Have a look at the revision number on server status page instead. It's only on newer versions, so it's not on LHC, and some projects with customized status pages don't have it (like SETI). However, by looking somewhere else I can tell you SETI has revision 13586 or higher.


OK understood.
Thanks PovAddict, but either way it needs to be updated.
ID: 17948 · Report as offensive     Reply Quote
bass4lhc

Send message
Joined: 28 Sep 04
Posts: 43
Credit: 249,962
RAC: 0
Message 17972 - Posted: 24 Sep 2007, 4:50:31 UTC

ok, i am not a forum person.
but changing the subject is not helping.

there was a situation where there was work which was not send. my first message.

then we go to boinc farms and something like "fair" placement of wu's. quota's,
even server updates.

please start your own thread.

@neasan&alex, this is your board, you have the power to keep it clean. i like to donate cpu time to lhc. i also would like to be taken a bit more seriously.

btw. i do not have a boinc farm.

ID: 17972 · Report as offensive     Reply Quote
picantecomputing

Send message
Joined: 1 Jan 07
Posts: 19
Credit: 44,943
RAC: 0
Message 17975 - Posted: 24 Sep 2007, 13:24:14 UTC - in response to Message 17972.  
Last modified: 24 Sep 2007, 13:25:09 UTC

ok, i am not a forum person.
but changing the subject is not helping.

there was a situation where there was work which was not send. my first message.

then we go to boinc farms and something like "fair" placement of wu's. quota's,
even server updates.

please start your own thread.

@neasan&alex, this is your board, you have the power to keep it clean. i like to donate cpu time to lhc. i also would like to be taken a bit more seriously.

btw. i do not have a boinc farm.

I'm not sure what your complaint is. You raised a concern, which Neasan answered here. Then the discussion evolved into a conversation on whether the admins' methods made sense. What else were you expecting to happen? Your concern was addressed, and then posters decided to explore the subject further. I don't think anyone failed to take you seriously - it's just that once your question was dealt with, the conversation moved on. BTW, this is pretty much how most threads go in this and other BOINC forums, so you shouldn't feel neglected. It's just kind of the way it goes. One idea leads to another, which leads to another, etc. That's the nature of discussion forums. If you don't feel you were taken seriously, you should probably re-read the thread, as it all pretty much relates to why there were WUs sitting there, ready to go, but not being sent.

Cheers.
ID: 17975 · Report as offensive     Reply Quote
uioped1

Send message
Joined: 26 Aug 05
Posts: 18
Credit: 37,965
RAC: 0
Message 17985 - Posted: 26 Sep 2007, 4:51:06 UTC - in response to Message 17924.  

Server SW upgrade would also allow "server side abortion", which would further reduce wasted CPU cycles.


Oh, so just because you think it's a good idea, now we all have to live with your immoral choices?

nevermind...
ID: 17985 · Report as offensive     Reply Quote
Speedy

Send message
Joined: 28 Jul 05
Posts: 37
Credit: 451,635
RAC: 20
Message 17988 - Posted: 27 Sep 2007, 6:52:56 UTC - in response to Message 17861.  

[quote]
Actually, I'm a little exited about the LHC switch-on this November ;)


Same here
Dose this mean that we will have a steady stream of work starting in November?

Have A Crunching Good day
ID: 17988 · Report as offensive     Reply Quote
Adam23

Send message
Joined: 30 May 06
Posts: 40
Credit: 219,288
RAC: 10
Message 17989 - Posted: 27 Sep 2007, 13:18:13 UTC - in response to Message 17988.  


Actually, I'm a little exited about the LHC switch-on this November ;)

Same here
Dose this mean that we will have a steady stream of work starting in November?


I don't think, they will switch it on this November. You better seek through this forum thoroughly, to find answers :)
ID: 17989 · Report as offensive     Reply Quote
bass4lhc

Send message
Joined: 28 Sep 04
Posts: 43
Credit: 249,962
RAC: 0
Message 17999 - Posted: 2 Oct 2007, 3:10:53 UTC - in response to Message 17975.  

ok, i am not a forum person.
but changing the subject is not helping.

there was a situation where there was work which was not send. my first message.

then we go to boinc farms and something like "fair" placement of wu's. quota's,
even server updates.

please start your own thread.

@neasan&alex, this is your board, you have the power to keep it clean. i like to donate cpu time to lhc. i also would like to be taken a bit more seriously.

btw. i do not have a boinc farm.

I'm not sure what your complaint is. You raised a concern, which Neasan answered here. Then the discussion evolved into a conversation on whether the admins' methods made sense. What else were you expecting to happen? Your concern was addressed, and then posters decided to explore the subject further. I don't think anyone failed to take you seriously - it's just that once your question was dealt with, the conversation moved on. BTW, this is pretty much how most threads go in this and other BOINC forums, so you shouldn't feel neglected. It's just kind of the way it goes. One idea leads to another, which leads to another, etc. That's the nature of discussion forums. If you don't feel you were taken seriously, you should probably re-read the thread, as it all pretty much relates to why there were WUs sitting there, ready to go, but not being sent.

Cheers.


dear picantecomputing,

my question/concern is not dealt with at all. there was a reply, not an answer. the explanation was (your favorite animal here)shit.

i don't mind discussions to evolve. i do mind not getting an answer and discussions that move away from a not dealt with topic.



ID: 17999 · Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : Number crunching : dear Neasan & Alex


©2024 CERN