Thread: Testing the cluster with symping

Symexec4.0 and symping4.0 comes pre-installed with the installation of Symphony DE 4.0 and after running the symphony configuration testing tool to test that all the components are working appropriately and responsive to workload, there seems to be an error coming up with regards to Session Manager saying "Failed to start session manager for consumer /SympTesting/Symping40. An operation timed out. Please make sure resources are available for session manager and try again".

I have a machine as the headnode (Management Node), with two windows (Compute host) and two Linux machines (Compute host).
Installation went pretty good as expected, I changed the configuration of the vem_resource.conf to ensure each compute host have the same configuration.

After successfully starting the relevant services on all hosts, I am only able to see the windows host started, but the two Linux says stopped and why does symping fail.
Has anyone come across this before? Really need your help on this.

Sounds like a problem in the vem_resource.conf file with the linux hosts.

Make sure the resource type is set correctly to LINUX86 and not NTX86

AGENT:8000:linuxhost:20:1:LINUX86:1

Please send me your vem_resources.conf file.

When you ran soamstartup on each host, make sure the process start_agent is running. There's a logs directory in each $SOAM_HOME/logs, so there may be more info there. Check the sd log on the master node, as it may have logged an error.

- Ajith

Hello Ajith,
Sorry about my late reply, the symping configuration tool is working now after following one of the instructions posted on this site, but the status of the windows hosts says started while the Linux systems say stopped despite starting them with the soamstartup$ command.
I have successfully pinged all the hosts, so there seems to be communication and the ports 15050 and 15051 are all available.
Is there any way you can help?
Cheers,
Mailfola1
Below this message is the current vem_resource configuration

---------------------------------------------------------------------------
#
# Resource configuration file
#
# For more details on configuring hosts for Symphony DE with vem_resource.conf,
# see the Symphony Reference in the Knowledge Center under
# "Configuration Files", vem_resource.conf.
#
# File format:
# <service name>:<port_number>:<host_name>:<Option1>:<Option2 >:<....>
# Valid service names are: AGENT, SD_SDK, SD_ADMIN, RS_DEPLOY, WEBGUI
#

#
# AGENT service information
# <service name>:<port_number>:<host_name>:<max number of SSMs SD can start>:<max number of SIMs SSM can start>:<OS_type>:<CPU_factor>

#MANAGEMENT HOST running SSM
AGENT:8000:tblood:50:0:NTX86:1


#COMPUTE HOSTS running SI and SIM
AGENT:8000:10.0.0.1:0:50:NTX86:1
AGENT:8000:bumblebee:0:50:NTX86:1
AGENT:8000:ratchet:0:50:NTX86:1
AGENT:8000:windcharger:0:50:LINUX86:1
AGENT:8000:starscream:0:50:LINUX86:1

#
# SD service information
# <service name>:<port_number>:<host_name>:<sd startcmd>
#
SD_SDK :15051:tblood:sd
SD_ADMIN:15050:tblood:sd

#
# RS service information
# <service name>:<port_number>:<host_name>:<rs startcmd>
#
RS_DEPLOY:15052:tblood:rs

#
# GUI service information
# <service name>:<port_number>:<host_name>:<gui startcmd>
# Note:The port number of WEBGUI is fixed as 18080.
#
WEBGUI:18080:tblood:startguiservice
---------------------------------------------------------------------------

Using Symphony DE on mixed windows / linux clusters is a bit inconsistent.

You can use the DE tray icon on Windows to stop the whole cluster. You can use the tray icon to restart only Windows systems.

The linux systems must be started using soamstartup on each machine. This command will start the start_agent process.

The tray icon should display the current status of all hosts in the cluster. If you suspect the status is not correct, ensure that the start_agent process is running on Linux. If start_agent is running and symping returns the status of the hosts, then there is something wrong with the status code in the tray icon.

- Ajith

Thanks for the prompt response, I started Symphony on both Linux systems without any problem, I checked the log and it started the agent on port 8000 on both hosts. I ran the command "start_agent" on them again to verify that the agent was running, the log then said
"ERROR [4613:3086661856] agent - Domain <System>: Start agent cannot listen on port 8000. Port may be in use. Check the port number defined in vem_resource.conf" that was a confirmation that the agent was running but the tray still says stopped for the LINUX systems after updating it.

If I am unable to get it to work, I might end up having seperate clusters for WINDOWS AND LINUX.

I am currently in search of the location of perf database, do I have access to it or it’s strictly for symphony without access.

What do you reckon I do?

Hi mailfola1,

I'm not at work, so I can't look at the code. If symping is reporting results from all your hosts, then I think your configuration is ok. Just use the cluster and ignore the tray icon status of the linux hosts. You can just use symping to get the status.

The only issue I can think of is if there's a DNS / hosts file issue between the tblood master and starscream and windcharger. To test, you can just run ping tblood from starscream and windcharger and ping startscream/ping windcharger from tblood. If you need to change a name to an ip-address or fully qualified domain name, make sure to update the vem_resources.conf file.

Normally the compute hosts will connect to the master host, so all hosts need to be able to create a connection on tblood.

Sym DE doesn't include perf or a database. Only the full version of Sym includes this feature.

- Ajith

Hello Ajith
Thank you very much for your prompt and detailed response. I will put your advice to practice now and hopefully get a good result.
Regards,
Mailfola1