Re: [Ganglia-developers] 2.5.6

[Federico S. <fd...@sd...>]

Matt et al,
I have some small changes that enable a "host_dmax" configure option 
for gmond. This allows you to tell gmond to remove a host after a 
period of inactivity. "If we do not hear from Host A for 10800 seconds 
(30 days), delete him from our local hash table." This feature is the 
complement to adding nodes, which gives some nice symmetry to gmond.

This is not a critical change, but we do see occasionally a dead node 
sticking around for weeks or months after it has been discarded from 
the cluster.

There is basically no new code besides the config-file option, since 
cleanup.c has had this ability for some time. I have tested it on 
Linux, and it works as expected. If you want this feature for 2.5.6, 
let me know and I will commit. Otherwise I will hold off until the next 
release.

-Federico

On Jan 15, 2004, at 10:21 AM, Brooks Davis wrote:

> On Thu, Jan 15, 2004 at 03:03:59PM +0800, Matt Massie wrote:
>> i'd like to release 2.5.6 sometime very soon.  let me know if anyone 
>> has
>> any objections.  brooks can you check that all the relevant freebsd
>> autoconf stuff is in the tarball?
>>
>> http://matt-massie.com/ganglia/
>>
>> i'm 99% sure it's all good because when i got your email i did a cvs
>> diff.. saw your changes.. cvs update and then make dist.
>
> It's not in there.  You've got configure.in 1.37 and you need 1.38 to
> get the removal of the bogus *ia64*) case at line 332.
>
> -- Brooks
>
> -- 
> Any statement of the form "X is the one, true Y" is FALSE.
> PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4
>
Federico

Rocks Cluster Group, San Diego Supercomputing Center, CA