[smokeping-users] Slave/Master Again
Jeff Terrell
jsterrel at unc.edu
Wed Oct 10 20:49:14 CEST 2007
On Oct 3, 2007, at 5:49 PM, Tobias Oetiker wrote:
>> Thank you Stephen and Tobi! With Stephen's change to my secrets
>> file, Tobi's
>> patch, and an overridden host-name with the -s switch, my master/
>> slave
>> architecture now works.
>
> is there anything you would add to the documentation so that it
> would be easier for others to set up ?
Haha, whoops, I take it back--the master/slave architecture doesn't
work for me after all. I was trying to add another slave (and
document the process well in order to help with the documentation).
The master is successfully pinging both of the slaves, which gives
some pretty graphs and made me think that the slaves were doing their
job. However, the slaves aren't pinging each other like I've told
them to.
I've noticed the following:
- tcpdump on the slaves tells me that there is, in fact, ICMP echo
requests and responses flowing in both directions between the hosts,
in bursts, just as I expect.
- Periodically, there is a TCP transfer from either slave to the
master. The slave transfers 853 bytes, and the master responds with
171 bytes.
- In the cache/data directory on the master, there are 4 files:
Artoo.rrd, Threepio.rrd, Artoo~threepio.rrd, and Threepio~artoo.rrd.
- The first two files have been updated within the past few minutes,
as shown by both the last-modified timestamp and a manual inspection
with `rrdtool fetch Artoo.rrd MAX -s 1191981600 -e 1192028400`.
- The last two files have not been modified for over a week. Both
`ls -l` and the rrdtool command above confirm this.
So, it would seem that there is a disconnect somewhere. As far as I
can tell, the slaves are doing everything they should, but for some
reason their data is not being added into the RR databases.
Thanks for all your help so far. I'm confident we'll get this
figured out soon! :-)
-Jeff T.
More information about the smokeping-users
mailing list