Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 1
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 2519 times and has 0 replies Next Thread
Sekerob
Ace Cruncher
Joined: Jul 24, 2005
Post Count: 20043
Status: Offline
Reply to this Post  Reply with Quote 
BOINC: "Update" button operations

"To Update or not to Update", that's the question!

First off, it does not update the BOINC client software. That is and will remain a manual event for the foreseeable future.

What it does is, to (try) make a live connection with the selected project(s),

  • To report completed work (See observation **!)
  • Retry a connection with the project servers when it previously failed, to for instance clear any results from the transfer queu.
  • Fetch any new tasks/work units if the cache/additional buffer queue has fallen below the desired minimum.
  • Start a sequence of connection attempts with ever increasing interval until there is a successful communication with the project servers.

    That latter feature essentially making repeated operation of the Update button pointless. BOINC will try again all by itself

    Mobile and part time crunchers will want to hit the Update button:

  • Before shutting down a device for the night or going on the road to force up any Ready to Report (RtR) entries. No use if you see in the Tasks window if there are no results "uploading" or in RtR state though.

    Note that BOINC will always try to upload queued result files immediately to keep the transfer window clear and take them off the client system for save keeping. This is a separate result store database from the very complex Scheduler which keeps track of who did what when.

    There are also technical reasons, to name a few:

  • Speed up transmission of changes made to the website device profiles to take immediate effect on a device/host. You don't have to as they will be anyhow on the next scheduled contact, e.g. with the reporting of results.
  • To check if a local network change made has not caused problems (e.g. IP address changes)
  • Your ISP took a longer holiday, router problems and the like.

    Whenever the client has diagnosed a network problem, it will start a retry-counter showing a deferral time that can be seen in the projects tab of the BOINC manager Advanced view. The servers will sometimes be on-line, but one or the other server daemon is paused or off line or plain very busy. Then the Project will instruct the client to back off. That can be 1, 5, 10, 60 minutes and more. It helps to ensure that when the project comes back on-line, the network does not get overloaded with thousands of simultaneous requests, almost as were there a Distributed Denial of Service attack.

    The message of all this is: Don't use the Update button if you don't have to. The less that button is pushed, the more efficient the grid can function, the more unique devices can be serviced in a unit of time

    thanks

    ** Tasks at WCG generally have 9 days deadline, sometimes less and for some more, meaning that even completed on the first day received, it's fine to not return the result until just before. Varying return dates have been calculated to allow great flexibility for the fastest and the slowest devices.

    == Related topics ==
  • Ready to Report
  • ----------------------------------------
    WCG Global & Research > Make Proposal Help: Start Here!
    Please help to make the Forums an enjoyable experience for All!
    ----------------------------------------
    [Edit 1 times, last edit by Sekerob at Mar 8, 2009 9:21:14 AM]
    [Dec 12, 2008 1:49:45 PM]   Link   Report threatening or abusive post: please login first  Go to top 
    [ Jump to Last Post ]
    Post new Thread