Just wondering what other sites working on 7.6.04.01 are seeing with regards to
pre-loading in the mid-tier.
Environment: mid-tier 7.6.04.01 on Tomcat 6.0.32 on JVM 1.6.0_24 x64 on Win2K3
R2 x64
- Pre-Production mid-tier has 2 quad-core i7s and 12 gb RAM and is running
several ARS-related apps as well
- Admin mid-tier is directly on the AR Server, which has 2 quad-core i7s and 24
gb RAM and is running several other ARS-related apps
Preload is running on both mid-tiers, for FQDN on the Admin box and both FQDN
and short name on the Pre-Prod box
(FQDN seems to be a requirement to make the Atrium Core Console work,
plus we prefer using it for DNS reasons)
Two problems that we are seeing appear to be related to the Preload process and
form caching:
1. Dialog boxes using Display Only Forms are VERY sluggish loading and saving.
The HPD:Help Desk Dialog form takes up to a minute to open and display, and
another minute to save and close (versus 8 seconds on our 7.1 mid-tier at the
very worst - but on 7.0.03 it is really a view of the actual HPD:Help Desk
regular form!). Opening CTM:People Search (another display only form) takes
20-30 seconds to open - enough that our support staff do not want to rely on it
to search for Login Name and want me to extend that particular required
customization back to HPD:Help Desk and now HPD:Help Desk Dialogs the way it is
in 7.0. Most of the other dialogs opened from the Quick Actions links have the
same problem - they take at least 20 or 30 seconds to load, versus response
times of no greater than 8 seconds for similar functions on our 7.1 mid-tier
(on MUCH older hardware). On that system, a comprehensive prefetch
configuration is the key to performance. The 7.6.04.01 preload for consoles
and regular forms seems to work fine, but for dialogs using display only forms
it is terrible. Our support staff testing the product do not think that this
is acceptable for production use.
2. Starting the tomcat (after the AR Server has started or has been running a
while), triggering a preload from persistent cache, has crashed the AREA plugin
thread several times now as it thrashes through all sorts of failures to
prefetch on various high-use forms (including those with poor response times)
on various usernames of people actively testing the system. The authentication
service then remains down until ARS is restarted, effectively blocking ALL
non-support staff from logging in to ITSM or Kinetic Request.
Related problems?; probably. Impact?; we will not go live until they are
solved. Just wondered what others were seeing out there... I have issues open
on both items.
Christopher Strauss, Ph.D.
Call Tracking Administration Manager
University of North Texas Computing & IT Center
http://itsm.unt.edu/
_______________________________________________________________________________
UNSUBSCRIBE or access ARSlist Archives at www.arslist.org
attend wwrug11 www.wwrug.com ARSList: "Where the Answers Are"
