Andrea,
Thank you for your response. Let me put it in a different way.
PostGIS Database Server has 45+ databases and tables(imported shapefiles) are
stored in default "public" schema. Each database has 8-10 tables. When I create
PostGIS connection to a database using GeoServer interface, I limit max
connections to that database as 8.
Based on your reply, I think it is where the connection pooling is enabled. Am
I correct?
Looks like JNDI doesn't help with my data organization.
I will have to restructure my data to put them in different schemas in the same
database. Correct?
Is there any script(s)/mechanism that help me to re-publish my data to make use
of JNDI?
I check the RAM usage using "Webmin" module where approx 2GB of RAM is given to
Java/Tomcat/GeoServer followed by the list of processes(sample provided below).
There is a significant amount of styling for each of the layers.
PID Owner Size Command
32426
pgsql 87580 kB postgres: postgres MyLayer1 127.0.0.1(33444) idle
"Size" is what telling me the RAM usage(Correct me if I am wrong). The
connections are made and you can see they become idle very soon after they are
made.May be they are sitting in the buffer pool of DB? I do not check the "HTTP
Caching" option while publishing the data layers.
Please let me know.
Thanks,
Ravi.
________________________________
From: Andrea Aime <[email protected]>
To: Ravi Pavuluri <[email protected]>
Cc: [email protected]
Sent: Thursday, April 21, 2011 2:45 AM
Subject: Re: [Geoserver-users] Geoserver Migration+ PostGIS(JNDI)
On Thu, Apr 21, 2011 at 12:38 AM, Ravi Pavuluri <[email protected]> wrote:
> Hi All,
>
> We are using 450+ vector layers(45+ databases) in PostGIS database served
> through GeoServer(2.0.2) most of which is WMS. Given the way it is, RAM is
> being used up ~60-100MB per layer(expected traffic is not very high as of
> now).We may hit the RAM limit if traffic becomes higher, since there is no
> queuing mechanism for total PostGIS connections . No such problem with RAM
> was encountered with shapefiles, though they are not recommended in
> production. We have not tried PostGIS (JNDI) but I read it that helps in
> connection pooling.
I don't follow... the PostGIS connection pool has a maximum configurable number
of connections. If you say the max is 20 every request trying to get
the 21th connection
will be queued.
That said, with 45 databases I guess you have a very large number if you sum
the max number of connections. Is it GeoServer or the postgis processes that
are consuming most of the memory?
I'm also puzzled as to how you computed that 60-100MB per layer memory
consumption.
The memory required by WMS is largely dependent on the number of requests going,
their size, and the styling used, it's not something statically depending on the
number of layers.
> 1) What is the best way to have all the databases from the "same" PostGIS
> schema to JNDI connections? Any estimate of how much(time) of an effort that
> would be?
JNDI helps if you have one database with 45 different schemas, if you have 45
phisically separate databases I don't believe you can setup a single connection
pool to talk with them.
> 2) Also, what would be the best way to migrate all of them to a stable
> version of GeoServer 2.1(whenever available)? Does combining JNDI
> connections and GeoServer 2.1 migration save time if we do them together?
>
> Any additional documentation on JNDI connection is appreciated.
We have a tutorial here:
http://docs.geoserver.org/2.0.0/user/tutorials/tomcat-jndi/tomcat-jndi.html
Cheers
Andrea
--
-------------------------------------------------------
Ing. Andrea Aime
GeoSolutions S.A.S.
Tech lead
Via Poggio alle Viti 1187
55054 Massarosa (LU)
Italy
phone: +39 0584 962313
fax: +39 0584 962313
http://www.geo-solutions.it
http://geo-solutions.blogspot.com/
http://www.youtube.com/user/GeoSolutionsIT
http://www.linkedin.com/in/andreaaime
http://twitter.com/geowolf
-------------------------------------------------------
------------------------------------------------------------------------------
Benefiting from Server Virtualization: Beyond Initial Workload
Consolidation -- Increasing the use of server virtualization is a top
priority.Virtualization can reduce costs, simplify management, and improve
application availability and disaster protection. Learn more about boosting
the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev
_______________________________________________
Geoserver-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-users