[RC5] Prediction updated for month of August.
wkitty42 at alltel.net
Mon Aug 2 19:14:46 EDT 2004
>> maybe my problem of them sitting on results is a config problem? the
>> client is set to use internal defaults instead of x results or y
>> time... i know that one of the machines flushed some 20+ results the
>> other day and i was normally running them so that they flushed when
>> they had 5 or so... and OGR is a different beast, altogether, in this
> run it with -config. 2, 10, and set it to 3 or 4 (usually I run it with
> -frequent anyway). You can also change 11 (buffer level check interval),
> every hour should theoretically be sufficient, depending on how large
> your buffers are.
i had thought about that but supposedly the client has internal defaults where you don't have to set "additional buffer checking"
what i used to do was to set the clients to flush the out buffers when there were one or two results in them but i got complaints
(indirectly) that that was causing additional bandwidth consumption on dnet's servers so i stopped it...
i seem to remember something about the clients having similar code to the perproxy in that they could determine what the best buffer
settings were based on how long it too them to process workunits and that they would adjust accordingly... i've noted that they seem
to be gathering ~20 workunits but i have seen times when there were more than 20 workunits in the out buffer and a full in buffer...
i definitely flushed them, then...
> Of course, if they're networked, you could just mount a network drive
> with the buffer files and get one client to do all the flushing.
they are but i can't do that for various reasons ;)
(@@) Waldo Kitty, Waldo's Place USA
__ooO_( )_Ooo_____________________ telnet://bbs.wpusa.dynip.com
_|_Eat_SPAM_to_email_me!_YUM!__|_____|_____ wkitty42 -at- alltel.net
More information about the rc5