Is anyone else experiencing lags in Spectrum 10.1? For example when I right click a device it take approximately 4 or 5 seconds for the pop up window to display. However, when I right click a Global Collection the pop up is nearly instantaneous.
Anyone else experiencing this issue or is it just us?
During beta nobody reported this lag and I don’t see one with my 10.1 SS in NY while my client is being launched near Boston. Do all of your clients have this problem?
Maybe it has something to do with java? If you try clearing your java cache and removing any old java on the client, does it help? Or are your java clients geographically far away from the tomcat server?
I’d definitely be interested to know if anyone else is seeing this…
We have 3 One Click servers. I am on the East Coast and the nearest to me is in Pacific Time Zone. Right clicking a Global Collection is instant, right clicking multiple devices is instant, right clicking from the Alarm tab is instant. Right clicking on device in the List tab is slower with times varying from 3 seconds to as many as 6 or 7 seconds. I tried from two other One Click Servers, one in Europe and another in Australia with similar results. Other team members are experiencing the same result.
I cleared my java cache including applications, downloaded the App again and get the same results. My current versions of Java are as follows:
Java 7 Update 55 (64 bit)
Java 7 Update 60
Java 8 Update 31 (64 bit)
Java 8 Update 66 (64 bit)
Should I consider removing Java and starting over?
OK, I decided to uninstall all instances of Java, reboot and re-install the version of Java from the Spectrum JRE page. No difference in performance.
I logged on to the One Click server and performed the same tests. Right clicking a device on the list tab takes about 1 sec. It's better but noticeably different than clicking an alarm, global collection or multiple devices on list tab.
Of the examples you used, I think the "device in the 'List' view" is the first example where a significant number of SNMP queries will be performed to populate the views. Something to consider.
The Global Collection example wouldn't issue any SNMP queries, and the alarm example I'm pretty certain has either No or very few SNMP queries as well.
Actually, that raises a good point. What if you right click your column headers and select the “Reset” at the bottom…Does the issue still exist?
I tried that. I un-selected all of them and only added back 3 or 4 fields. Same results.
I did some tests for this too - and allow me this idea:
- in case the device is "not responding to polls" (I just changed the network address) this would indicate the SNMP-poll (positive or negative / timeout) is not relevant.
it appears to me, it is not relevant - I do not see a response-time difference
- adding "integrations" - like eHealth or "CA PC" (as by the right_clickt the device models context is verified) - having them "alive"/or stopped - may affect the right_click.
So - it remains the context for the "native topology" - and here I expect the number of models in the container (when in the topology view) may affect this. I do not see this at a serious level - but I expect this should be checked.
The other thing is the OC-server performance itself - the indicator for this is "Test_1" when opening in OC-Console / Help / Debug Console -> Console Test "Run Test". have look to Ranking OC-Console/Workstation to OC-Server performance
I feel there is a rendering active when using the Topology View - which then affects the right_click. I.e. having very high number of connection within the Container / or "going out" via OffPageReferences. May you can verify this - and moving the device into a different container which is with less workload - and if this then siginificantly improves the "right_click".
Your OneClick servers/client have enough memory? sometimes I've seen in the past that when the client or server doesn't have enough memory to process Java requests, it tends to create a lag for some actions within OneClick. it doesn't hurt to check it out
Configure OneClick Client Memory Settings - CA Spectrum - 9.4.1 to 9.4.3 - CA Technologies Documentation
Configure OneClick Web Server Memory Settings - CA Spectrum - 10.0 - CA Technologies Documentation
good info however I'm not sure it will help us. Here's what I have.
Windows server has 16GB of memory with 11GB available.
One Click Webserver is allocated 768MB and is using 353MB (52%)
One Click Java Memory usage: Min client memory usage is 1024MB and Max is 2048MB
While I don’t think this is the cause, your 10.1 OC tomcat service should be configured for 4GB, not 768M…
One thing that has helped a couple of customers is to edit the $SPECROOT\tomcat\webapps\spectrum\WEB-INF\web.xml and change this to 5000:
/custom area. Sometimes we’ve seen an old customization that worked in a previous version need an update for the current version.
can you tell me which line to look at in the web.xml file?
My apologies, I don’t know why half of my content gets ripped out of my responses. Change the “com.aprisma.topo.ModelRepository.ThrottledModelReadCount” to 5000 and stop/restart tomcat:
I'm always on the look-out for optimization tips. Is this something you would recommend applying to all environments, even if ours isn't necessarily feeling the same lag as reported in this thread?
I've done some more research. Here's some more information.
Issue: One Click seems slow for me and my co-workers. One Click is slow to respond when right clicking devices in the List view. Also it takes 8-10 minutes for me to open the Event Configurator.
I went ahead and opened a ticket with support.
We have Spectrum 10.0 in production, and the Event Configuration will hang for a period of minutes while retrieving events. If Spectrum 10.1 has the same issue with EC, then it's a defect that hasn't yet been fixed.
I also have the same problem of freeze when I am connected to the Spectrum console (version 10.1).Does anyone have a solution to fix this very embarrassing problem?