perfsonar development work

[Maciej Glowiak <>]

ulisses wrote:
> Hi all
>
> I wanted to know for how long a register or a keepalive message makes the 
> data valid
>
> looking at
>
>         http://wiki.perfsonar.net/jra1-wiki/index.php/Lookup_Service_resources
>
> I haven't found this information. 
>
> I was interesting becuase in the topology service I was planning a similar 
> register procedure.
>
> Maybe the register/keepalive __response__ should include the number of 
> seconds for which the data
> is garateed it will be stored?

Hi Ulisses,

I've just commited changes in LSCleanup component which is responsible 
for removing old, obsolete Lookup Information from the database.

Some more info is provided on 
https://bugzilla.perfsonar.net/show_bug.cgi?id=67

but basically what you must do to enable LSCleanup is

1. Enable components: Scheduler and LSCleanupLoader in 
components.properties (may be done by new "ant configure-xmlls"):

--------------------------------------------------------------
# LS Cleanup (loads by LSCleanupLoader, needs scheduler)
ls_scheduler=org.perfsonar.commons.auxiliary.components.simplescheduler.Scheduler
ls_cleanup_loader=org.perfsonar.service.lookupService.xmlType.LSCleanupLoader
--------------------------------------------------------------

2. Set configuration of LSCleanup in service.properties file:

--------------------------------------------------------------
#LS Cleanup loader
component.ls_cleanup_loader.interval=30
component.ls_cleanup_loader.scheduler_component=ls_scheduler
component.ls_cleanup_loader.storage_component=ls_storage
--------------------------------------------------------------

3. Set TTL for Lookup Information (now it's common for all Lookup 
Information and the LS decides how long it'll keep lookup info)
- it should be also done in service.properties

--------------------------------------------------------------
service.ls.ttl=60
--------------------------------------------------------------

So, component.ls_cleanup_loader.interval=30 tells us that Scheduler will 
be launching LSCleanup action each 30 seconds and "service.ls.ttl=60" 
means that LSCleanup will be removing data older than 60 seconds.

Of course these values are for testing ONLY. I think "TTL" should be 
higher than 1 hour (3600 sec) and then the "interval" may be 10 minutes 
(600 seconds)  -- it all depends how much services we have in LS and how 
often they register to our LS.

In some cases TTL may be 1 day and then the "interval" may be 1 hour. It 
all depends.

Best regards

Maciej

--

 --------------------------------------------------------------------
| Maciej Glowiak        Network Research and Development            ||
| 

     Poznan Supercomputing and Networking Center ||
| (+48 61) 858 2024     http://monstera.man.poznan.pl/              ||
 ====================================================================