[Bamboo-Users] Get time on Grid5000 cluster
Sean Rhea
sean.c.rhea at gmail.com
Tue Oct 2 21:19:40 UTC 2007
The default retry time for gets in Bamboo is 5000 milliseconds:
Dht.java:1075: get_timeout = configGetInt(config, "get_timeout", 5000);
That should explain the behavior you're seeing. I'm not sure why
they're taking so long, though. Do you have nodes dying a lot? Is
the network congested?
Sean
On 9/27/07, Martina.Baldanzi at isti.cnr.it <Martina.Baldanzi at isti.cnr.it> wrote:
> Dear All,
> I'm Martina. I have a question about running Bamboo on a cluster
> (Grid'5000 https://www.grid5000.fr/mediawiki/index.php/Grid5000:Home) .
> I have to realize a Resource Discovery Service using Bamboo, but first
> of all, I have to measure the Get response time of Bamboo. After this
> testing, I obtain strange results, or better, results that I don't
> understand. I explain better the problem:
> - For testing Bamboo, I use the Rennes site of Grid5000 and I have
> reserved 32 nodes (32 random nodes from Rennes site).
> - Starts from frontend.rennes.grid5000.fr, for each node N in 32 node
> list, I open an ssh connection and N runs a Bamboo instance.
> - Starts from frontend.rennes.grid5000.fr, for each node N in 32 node
> list, I open an ssh connection and every node N runs only one Put to
> publish its computational resource info. The key is a string (ip+port of
> node N) and the value is the content of an XML file with description of
> N computational resource (dim ~800 bytes)
> - Starts from frontend.rennes.grid5000.fr, I choose one node X in 32 node
> list and I open an ssh connection and X runs a Daemon that periodically
> invokes the bamboo Get class. The key is the string ip+port of one node
> randomly choosen from 32 list node.
>
> In the first test, the daemon invokes the get class every 11 sec, in the
> second test, the daemon invokes the get class every 27 sec.
>
> The results are (Bamboo get response time):
> Period: 11000 ms
> Milliseconds: 147
> Milliseconds: 18
> Milliseconds: 12
> Milliseconds: 7
> Milliseconds: 9
> Milliseconds: 5012
> Milliseconds: 5035
> Milliseconds: 6
> Milliseconds: 5
> Milliseconds: 5039
> Milliseconds: 10
> Milliseconds: 10
> Milliseconds: 2
> Milliseconds: 8
> Milliseconds: 5035
> Milliseconds: 7
> Milliseconds: 3
> Milliseconds: 3
> Milliseconds: 9
> Milliseconds: 4
> Milliseconds: 5019
> Milliseconds: 4
> Milliseconds: 5
> Milliseconds: 3
> Milliseconds: 5
> Milliseconds: 5022
> Milliseconds: 3
> Milliseconds: 4
> Milliseconds: 24
> Milliseconds: 3
>
> Period: 27000 ms
> Milliseconds: 5
> Milliseconds: 30
> Milliseconds: 4
> Milliseconds: 13
> Milliseconds: 16
> Milliseconds: 20
> Milliseconds: 3
> Milliseconds: 4
> Milliseconds: 4
> Milliseconds: 3
> Milliseconds: 4
> Milliseconds: 3
> Milliseconds: 5
> Milliseconds: 4
> Milliseconds: 4
> Milliseconds: 5007
> Milliseconds: 5
> Milliseconds: 3
> Milliseconds: 2
> Milliseconds: 10018
> Milliseconds: 3
> Milliseconds: 4
> Milliseconds: 3
> Milliseconds: 5010
> Milliseconds: 5
> Milliseconds: 8
> Milliseconds: 5016
> Milliseconds: 5013
> Milliseconds: 2
> Milliseconds: 3
> Milliseconds: 5011
>
> We can see that in some cases I obtain good response time 2, 3 ms, while
> in other cases I obtain 5000 or 10000 ms.
> Have you some ideas about these strange results?
> Help me please.
> Thank you very much,
> Martina
>
> ps. In the attached file there is an example of config file
>
> _______________________________________________
> Bamboo-Users mailing list
> Bamboo-Users at bamboo-dht.org
> http://bamboo-dht.org/cgi-bin/mailman/listinfo/bamboo-users
>
>
>
--
"Humanity has advanced, when it has advanced, not because it has been
sober, responsible, and cautious, but because it has been playful,
rebellious, and immature." -- Tom Robbins
More information about the Bamboo-Users
mailing list