guys- http://matt-massie.com/ganglia/ganglia-2.6.0.200501251557.tar.gz
is a new snapshot that fixed the deaf/mute bug that causes gmond to shoot to 100% CPU use... see...
http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=9this snapshot also added a new attribute "timeout" to the tcp_accept_channel to hopefully fix the following bugs...
http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=7 http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=8 .. submitted by josh.the new timeout attribute allows you to specify a timeout in microseconds for sending data to client that connects to the tcp_accept port. for example...
tcp_accept_channel {
port = 8649
timeout = 1000000
}
...would set a timeout of 1 sec for writing to clients. in previous
snapshots if the client blocked the server would close the connection
and not try again.
specifying a timeout causes the server to wait timeout microseconds before closing the connection. keep in mind this does not mean that the _entire_ xml has to be sent in "timeout" time (which might be impossible if you have a slow link or a HUGE cluster like josh). this timeout applies to each small write.
josh, please let me know if this fixes your problem. oh the problem we have when we have a HUGE cluster http://www.tcf.vt.edu/. :) very nice apple cluster...
...speaking of apple.. i have not dropped in the new metrics that eric wages (an apple developer) submitted to the list today. i'll work on that next.
yemi, your network metrics are in the basic configuration for solaris now.
please submit bug reports to bugzilla.ganglia.info.
why did the sourceforge cvs server choose this week to slow to crawl?
-matt
--
PGP fingerprint 'A7C2 3C2F 8445 AD3C 135E F40B 242A 5984 ACBC 91D3'
They that can give up essential liberty to obtain a little
temporary safety deserve neither liberty nor safety.
--Benjamin Franklin, Historical Review of Pennsylvania, 1759
signature.asc
Description: OpenPGP digital signature
