We have upgraded a few months ago to the Prizm 3.0 running on CentOS. Overall the software works much better than the old versions.
I use remote desktop app to access the prizm server on the local machine.
When ever I open the ‘Define networks’ page, Prizm is slow, every click takes 5-10 minutes to refresh. I have turned off ‘auto discovery’ with the same results. This is quite a challenge for both administrator and installers when installers are in the field trying to add a new canopy sm, as they have to wait for about a 1/2 hour for the radio to be discovered and accepted to BAM.
Is there anything I can do to speed things up ?
so i found part of it being that the SNMP settings on selected radios was set to READ ONLY, the other thing I found was that two sm’s had identical ip management addresses.
What database are you using? If you look at the About Box -> Advanced Tab, it will tell you how long the db queries are taking. Do you see high numbers on that?
What would be considered high numbers on the DB stats do you think?
I would expect most queries to be less than a second or so to execute. Certainly less than 10 seconds.
If you are seeing queries averaging in the 10s or 100s of seconds, then there’s some cleaning up/optimization that probably needs to be done.
Since most User activities requires some level of db queries, that is typically where slow responsiveness would be perceived.
Typical stats we look for is and publish in the dialog box, how long do we wait to get a connection, and how long do the queries take on average. If the system is waiting too long, then we’d want to understand why, and tune it appropriately.
Ok.
That’s good information.
Looks like my query times may be a bit high.
Here are my stats on Prizm 3.1Beta
DatabaseManager
Average Connection Wait Time = 0
Average Query Execution Time = 7
Connection Hold Time = 300000
Connection Timeout Count = 0
Connection Wait Time = 30000
DB Connections Opened = 7
Database Profile = sqlsvr.xml
IP Address = 192.168.10.5
JDBC Driver = SqlServer
JDBC Jar File = sqljdbc.jar
Max. Connections = 50
Port = 1433
Schema Name = prizm3
TaskScheduler
No. of scheduled Tasks = 24
NodeManager
Event Queue Size = 0, 0, 0, 0, 13, 14, 14, 14
No. of registered Listeners = 8
Processing Time =
update (blocking): 318
refresh (blocking): 1934
up (non-blocking): 2134
down (non-blocking): 10025
update (non-blocking): 5909
refresh (non-blocking): 30167
link_update (non-blocking): 7060
SNMPLib
Trap Port = 162
ElementConfigManager
Current Job =
Current Job ID = 0
Num Change Jobs Queued = 0
Num Config/Update Jobs Queued = 0
NorthboundInterface
SNMP Agent Enabled = true
Soap/XML Enabled = true
Average Connection Wait Time = 0
Average Query Execution Time = 7
Connection Timeout Count = 0
DB Connections Opened = 7
The significant information from the set as above does not seem to indicate that the database is the bottleneck. (Time is in millis).
Do you run the client on the same machine as the server?
Question: Is your database on the same server as the Prizm install?? Reason I ask is the IP address listed in your post. Mine shows 127.0.0.1 and I have no waiting queues normally (unless there are several people working in Prizm at the same time).
If the DB is on the same machine you should address and talk to it via the localhost (127.0.0.1) which does not require actual network activity. If you are talking to the network, that might be much of your response time issue.
–
Larry Smith
SysAd ECSIS.NET
sysad@ecsis.net
Yes the Database resides on the same server. I notice some of the numbers are substantially higher than that of the previous post.
I went through all 86 radios we have to find about 8 that did not have the region code set (due to recent updates to 8.2.2). I am using a /22 (1022 hosts) for the network after scaling back from a /16, in hopes of speeding up network speeds if that what was causing it. I manually added ip’s to 8 sm’s last night and accepted to BAM, it took 4 hours to search those ips, and started to accept 9.30pm. It completed at 6:05am this morning. 1 hour per sm. Am I the only one having these issues, with so little radios. Our entire network is canopy, no generics.
Here is the info:
DatabaseManager
Average Connection Wait Time = 0
Average Query Execution Time = 0
Connection Hold Time = 300000
Connection Timeout Count = 0
Connection Wait Time = 30000
DB Connections Opened = 9
Database Profile = postgresql.xml
IP Address = 127.0.0.1
JDBC Driver = PostGreSQL
JDBC Jar File = pg74.214.jdbc3.jar
Max. Connections = 50
Port = 5432
Schema Name = prizm
TaskScheduler
No. of scheduled Tasks = 23
NodeManager
Event Queue Size = 0, 0, 0, 0, 0, 0
No. of registered Listeners = 6
Processing Time =
delete (blocking): 929
update (blocking): 276
refresh (blocking): 417
update_ipmac (blocking): 1124
up (non-blocking): 2714857
down (non-blocking): 2628059
update (non-blocking): 171
refresh (non-blocking): 11640864
add (non-blocking): 14183636
SNMPLib
Trap Port = 162
ElementConfigManager
Current Job =
Current Job ID = 0
Num Change Jobs Queued = 0
Num Config/Update Jobs Queued = 0
BAMManager
Radius Enabled = false
Prizm 3
I am having the same problems with Prizm.
It takes forever to to do anything, even move between views.
I no longer hold up my installation crews while trying to accept new SM’s.
On the Server in task manager, I will see 1- 15 postgres processes running at the same time.
When the app locks up, I can see 4 of the postgres processes taking between 20%- 25% of the CPU each.
Prizm has always been slow since we installed it about two years ago however, now it borders on insanity.
Any help would be much appreciated!!
BELOW, are the Advance tab stats
DatabaseManager
Average Connection Wait Time = 0
Average Query Execution Time = 7
Connection Hold Time = 300000
Connection Timeout Count = 0
Connection Wait Time = 30000
DB Connections Opened = 12
Database Profile = postgresql.xml
IP Address = 127.0.0.1
JDBC Driver = PostGreSQL
JDBC Jar File = pg74.214.jdbc3.jar
Max. Connections = 75
Port = 5432
Schema Name = Sylacauga Canopy
TaskScheduler
No. of scheduled Tasks = 54
NodeManager
Event Queue Size = 0, 415, 416, 416, 416, 416
No. of registered Listeners = 6
Processing Time =
delete (blocking): 1107
update (blocking): 29
refresh (blocking): 3571
up (non-blocking): 850159
down (non-blocking): 859777
update (non-blocking): 431613
refresh (non-blocking): 612351
add (non-blocking): 1352878
link_update (non-blocking): 700636
SNMPLib
Trap Port = 162
ElementConfigManager
Current Job =
Current Job ID = 0
Num Change Jobs Queued = 0
Num Config/Update Jobs Queued = 0
Thanks very much,
Reay
After performing a vaccume on the database the results and speeds improved substantially:
DatabaseManager
Average Connection Wait Time = 0
Average Query Execution Time = 0
Connection Hold Time = 300000
Connection Timeout Count = 0
Connection Wait Time = 30000
DB Connections Opened = 5
Database Profile = postgresql.xml
IP Address = 127.0.0.1
JDBC Driver = PostGreSQL
JDBC Jar File = pg74.214.jdbc3.jar
Max. Connections = 50
Port = 5432
Schema Name = prizm
TaskScheduler
No. of scheduled Tasks = 22
NodeManager
Event Queue Size = 0, 0, 0, 0, 0, 0
No. of registered Listeners = 6
Processing Time =
update (blocking): 17
up (non-blocking): 595
down (non-blocking): 3472
update (non-blocking): 82
refresh (non-blocking): 18637
SNMPLib
Trap Port = 162
ElementConfigManager
Current Job =
Current Job ID = 0
Num Change Jobs Queued = 0
Num Config/Update Jobs Queued = 0
BAMManager
Radius Enabled = false
What did you do to your database?
I am having the same slow issues.
Hey any help would be greatly appreciated, I am about to wipe everything I have done clear and start with linux red hat. I have Prizm 3.0 running on CentOS and cannot make the sm’s authenticate or any BAM plans to work. Prizm runs fine otherwise, updates software and everything. Am I missing something? If I set the AP’s to authenticate and put in the Prizm server’s IP address nobody connects to them. I have the server running on the network with a local IP address (10.1.1.1-10.1.2.100) the same as my AP’s and SM’s (the only thing on that server is Prizm 3.0). Here is what mine say s in the about box under server and advanced. Again any help would be greatly appreciated, I don’t want to do this all over again.
Features: Full (EMS, BAM)
Java Version: 1.5.0_08
Platform: i386
Operating System: Linux
OS Version: 2.6.18-53.1.4.el5
Server Time: 2/25/08 10:09 AM
Up Time: 012:22:35:22
Maximum Memory: 493M
Total Memory: 166M
Free Memory: 108M
Total Threads: 114
Server IP: 10.1.1.2
DatabaseManager
Average Connection Wait Time = 0
Average Query Execution Time = 1
Connection Hold Time = 300000
Connection Timeout Count = 0
Connection Wait Time = 30000
DB Connections Opened = 2
Database Profile = mysql.xml
IP Address = 127.0.0.1
JDBC Driver = MySql
JDBC Jar File = mysql-connector-java-3.0.17-ga-bin.jar
Max. Connections = 50
Port = 3306
Schema Name = prizm
TaskScheduler
No. of scheduled Tasks = 21
NodeManager
Event Queue Size = 0, 0, 0, 0, 0, 0
No. of registered Listeners = 6
Processing Time =
delete (blocking): 58
update (blocking): 25
refresh (blocking): 968
up (non-blocking): 133
down (non-blocking): 144
update (non-blocking): 131
refresh (non-blocking): 20303
add (non-blocking): 95
link_update (non-blocking): 1
SNMPLib
Trap Port = 162
ElementConfigManager
Current Job =
Current Job ID = 0
Num Change Jobs Queued = 0
Num Config/Update Jobs Queued = 0
BAMManager
Radius Enabled = false
IF THIS MEANS ANYTHING TO YOU!!
Vacuume the database worked
we have it scheduled now to do this every hour, and seems to speed up Prizm performance substancially.
vacuumdb -fv -d prizm -U bost_user -H