[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
RE: [eclipselink-users] RMI Cache Performance Problems
|
I can not think of any reason for the cache coordination to affect the lifecycle of the EclipseLink Session. I would recommend tracking down why the Session is logging out and prevent it as Session startup can be expensive.
--Gordon
-----Original Message-----
From: eclipselink-users-bounces@xxxxxxxxxxx
[mailto:eclipselink-users-bounces@xxxxxxxxxxx]On Behalf Of Vespa,
Anthony J
Sent: Monday, December 10, 2007 10:06 AM
To: EclipseLink User Discussions
Subject: RE: [eclipselink-users] RMI Cache Performance Problems
My code restructure seemed to fix all the issues I was having - though
it does seem that the sessions still do disconnect after 30 seconds of
inactivity (probably due to garbage collection) - though this didn't
seem to happen before I implemented the RMI calls.
I am doing some simple queries returning anywhere from 1-20 rows of data
with simple joins. I set up all my objects to lazy load and do cache
invalidate.
-----Original Message-----
From: eclipselink-users-bounces@xxxxxxxxxxx
[mailto:eclipselink-users-bounces@xxxxxxxxxxx] On Behalf Of Gordon Yorke
Sent: Monday, December 10, 2007 9:59 AM
To: EclipseLink User Discussions
Subject: RE: [eclipselink-users] RMI Cache Performance Problems
Cache Coordination should have little impact on read performance. It
may cause TopLink to read from the database more often but those reads
should not be any slower than the original reads to load the data. Are
you using indirection or Lazy loading? How much data is being read by
the query? Is there a lot of coordination traffic? What level of
coordination are you using invalidation or full changesets (default)?
--Gordon
-----Original Message-----
From: eclipselink-users-bounces@xxxxxxxxxxx
[mailto:eclipselink-users-bounces@xxxxxxxxxxx]On Behalf Of Vespa,
Anthony J
Sent: Friday, December 07, 2007 3:22 PM
To: EclipseLink User Discussions
Subject: RE: [eclipselink-users] RMI Cache Performance Problems
Thank you for filing the issue. :)
I have logging on full, and the latency occurs on loads and only after I
add the cache coordination. The exact same load call seems to take 5
seconds under coordination as opposed to less than one sec. Since it is
a call to something already cached, sometimes it just waits as it pulls
it back from the cache, and sometimes it hits the SQL, that goes through
quickly, and the return takes another 3-4 secs.
-----Original Message-----
From: eclipselink-users-bounces@xxxxxxxxxxx
[mailto:eclipselink-users-bounces@xxxxxxxxxxx] On Behalf Of Gordon Yorke
Sent: Friday, December 07, 2007 10:20 AM
To: EclipseLink User Discussions
Subject: RE: [eclipselink-users] RMI Cache Performance Problems
Hello Tony,
1) This may be a classloader issue, I've filed a bug
https://bugs.eclipse.org/bugs/show_bug.cgi?id=212270 In the bug I have
attached a potential patch can you let me know if the patch resolves
this problem?
2) Are these performance issues occuring only once you add the cache
coordination? What operations do the calls consist of? Reads, writes,
combinations? How to the operations differ between slower calls and
faster calls? Have you turned on logging? Does the call seem to wait on
the database or is it after the writes to the database have completed?
--Gordon
-----Original Message-----
From: eclipselink-users-bounces@xxxxxxxxxxx
[mailto:eclipselink-users-bounces@xxxxxxxxxxx]On Behalf Of Vespa,
Anthony J
Sent: Thursday, December 06, 2007 4:45 PM
To: EclipseLink User Discussions
Subject: [eclipselink-users] RMI Cache Performance Problems
So I've got RMI coordinated caching working across two servers and it
behaves as I would expect. I've got questions:
1) I seem to be getting this error a lot:
ava.util.MissingResourceException: Can't find bundle for base name
org.eclipse.persistence.exceptions.i18n.CommunicationExceptionResource,
locale en_US I assume some sort of missing resource for different
languages is somewhere, either I forgot it or it wasn't included.
2) I am having some performance issues, to the effect of like 10-15
seconds per call, where before they took less than a sec to 2 seconds.
In some cases the connections are being closed after a very short
(period of seconds) inactivity, and sometimes it seems to just hang
before returning data. I'm wondering if I am missing some obvious
settings or something? I don't close the Entity managers after each web
service request so I'm trying to configure the JDBC timeout but that
doesn't seem to have any effect.
Thanks!
T
_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users
_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users
_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users
_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users
_______________________________________________
eclipselink-users mailing list
eclipselink-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipselink-users
