[rc5] v2 client fast but crappy

Jeff Lawson jlawson at hmc.edu
Sun Jun 29 03:44:45 EDT 1997


At 09:26 AM 6/29/97 +0200, Honza Pazdziora wrote:
>Unfortunately, in the docs there is no mention about where the status
>is saved. I have 80 machines (Un*ces and Windoze), most Un*ces are
>mounting the directory via NFS.
>
>Why not add feature similar to that in Apache: instead of saying how
>much blocks it should download, tell the client how much spare blocks
>it should keep. I have a permanent net connection but sometimes the
>clients just fill the log with Network Errors. If I could specify to
>always try to have let's say 2 more blocks, they could overcome some
>60 minutes without sitting idle.

The v2 clients are designed to allow sharing of the buffer files by
multiple processes.  By this manner, we can easily support "multi
threading" equally easily on all platforms.  Thus, you would simply run
"rc5v2 -fetch" and "rc5v2 -flush" in a repeating script (that has a sleep
delay in it) in the background of the main client.  This will ensure that
the main client always has disk buffers to work from and it will not even
attempt to do network communications at all.


Jeff
----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list