cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Product Manager
Product Manager

Linux Agent Deployment

Linux Agent Deployment

Add Node Wizard - Push Deployment

Deploying the Linux Agent to an individual machine is as simple as adding the node to Orion via the Add Node Wizard. To begin, navigate to [Settings -> All Settings -> Add Node], enter the IP address or fully qualified host name of the Linux host you'd like managed in the "Polling Hostname or IP Address" field, and select the "Windows Servers: Agent" radio button from the available "Polling Method" options. Next, enter the credentials that will be used to both connect to the Linux host and install the agent software. The credentials provided here should have 'root' or equivalent level permissions. Note that the credentials provided here are used only for initial deployment of the agent. Future password changes of the account credentials provided here will have no impact on the agent once it is deployed.

The Agent is deployed to the Linux host using a combination of SSH and SFTP requiring TCP port 22 be open from the Orion server (or additional polling engine) to the Linux endpoint you wish to manage for push deployment to function properly.

Agent Deployment Add Node.png

Once credentials are provided click "Next" at the bottom of the page. You will then be prompted to start the installation of the Linux Agent. Click "Start Install" and the progress indicator will appear. When finished, you will be taken through the rest of the Add Node Wizard flow where you can select which resources you wish to monitoring on the host.

Install Agent PromptInstalling Agent Progress IndicatorList Resources
Install Agent Software.pngInstalling Agent Software.pngList Resources.png

Manual - Pull Deployment

In some scenarios it may not be possible for the Orion server to push the agent to the Linux host over SSH. This is not uncommon when the host you wish to manage resides behind a NAT or is hosted in the cloud. While firewall policy changes, port forwarding, or one-to-one address translations could be made to facilitate push deployment of the agent, in many cases it may be far easier to perform a manual deployment of the agent to those hosts.

The Linux Agent can be downloaded from the Orion web interface to the Linux host by going to [Settings -> All Settings -> Agent Settings -> Download Agent Software] and selecting "Linux" from the options provided and clicking "Next". In the following step of the Wizard select "Manual Install" and click "Next". Finally, In the third and final step of the wizard is where you will select the Linux distribution you will be installing the Agent on, as well as bitness of the OS (32 or 64bit). Here you can also configure any advanced options the agent will use when it is installed, such as which polling engine the Agent should be associated with in Agent Initiated (Active) mode, or the listening port the Agent will use when running in Server Initiated (Passive) mode.

Select Agent TypeSelect Deployment MethodChoose Agent Settings & Distribution
Select Agent Type.pngAgent Deployement Type.pngAgent Distribution.png

Once selecting all the appropriate configuration options, click the "Generate Command" button at the bottom of the page. This will generate a dynamic installation command based upon the the settings chosen above, which can then be copied and pasted into an SSH or X-Windows session on the Linux host. The Linux machine will then download and install the appropriate agent software from the Orion server using those pre-configured options.

Copy Install Command.png

Paste the generated command into your Linux terminal session and press 'enter' to start the download and install process.

Console Session Install Agent.pngThe command will begin downloading, then installing the Agent onto the machine. When complete, the agent service will automatically start on the Linux host registering with the Orion server, and becoming a managed node.

Agent Download and Install.pngOnce registered select your newly added agent node and click "Choose Resources" from the 'Manage Agents' view to select items on the node you would like to monitor.

Choose Resources.png

Mass Deployment - Repository

With the introduction of the Linux Agent in SAM 6.3 Beta 3, the Orion server and Additional Web Servers are now also Linux repositories for the Agent. This method can be utilized for mass Agent deployment using automation and orchestration tools such as Puppet and Chef. It also means you can use the same native built-in  package management tools for the Agent, as you would any other package on the Linux operating system. Now you can install the Linux Agent the same way you might install or update BIND or Apache, using 'yum', 'apt-get', or 'zypper' depending upon your distribution. To utilize 'yum', 'apt-get', or 'zypper' for installing the agent you must first register the Orion repository with your operating system. To do so, navigate to [Settings -> All Settings -> Agent Settings -> Download Agent Software -> Linux] and select 'Install via Package Management Tool (e.g. yum, apt)'.

yum.pngBased upon the distribution selected in the drop down (1), a command line string is dynamically generated. Simply click the 'Copy' button (2), then paste this string into an SSH session for the Linux host you want to add the repository to and press 'Enter'. Once the repository is added to the host, you can use the appropriate 'yum', 'apt-get', or  'zypper' command to install the agent from the repository.  Simply click the 'Copy' button (3) and once again, paste this command into the SSH session to install the Agent. The Agent will then be downloaded from the Orion repository and installed on the host.

Select Distribution
Add Repository to Host
Linux Repository.png

SSH Add Repository.png

Install AgentConfigure Agent

CopyPaste Command From Step #3 Above

yum install.png

IP/Hostname, Username & Password

Configure Agent.png

When installation is complete, all that remains is to configure the agent by typing 'service swiagentd init'. This will bring up the Linux Agent configuration settings where you will need to provide the IP, hostname, or FQDN of the Orion server or Additional Polling Engine (option #2). Then enter your Orion 'admin' or equivalent credentials used to login to the Orion web interface (Options 4 & 5). When complete, save your changes and exit (Option 7). This will then register the agent with the Orion server and begin managing it as a node.

Add Node WizardPull Deployment
Deployment via Repository Manual Installation
78 Replies
Level 16

@aLTeReGo  we are facing one strange issue for 1 Linux server. Tried to install the agent from Solarwinds console, but it doesnt go thru and states that agent installation could not be completed due to unknown reasons... Then we tried manually from the Linux VM end but there also it wasnt allowing so we removed the agent and thought of re-installing...

Now the agent is installed but service is not starting on it. If we give service swiagent init it does show options to enter but at the end it just fails stating either service is not able to start or there is no end point to resolve...

0 Kudos


@pratikmehta003 wrote:

@aLTeReGo  we are facing one strange issue for 1 Linux server. Tried to install the agent from Solarwinds console, but it doesnt go thru and states that agent installation could not be completed due to unknown reasons... Then we tried manually from the Linux VM end but there also it wasnt allowing so we removed the agent and thought of re-installing...

Now the agent is installed but service is not starting on it. If we give service swiagent init it does show options to enter but at the end it just fails stating either service is not able to start or there is no end point to resolve...


After installing the Agent did you configure it? Are you using Active or Passive Agent mode? From what it sounds like, there may be something blocking traffic from the Agent to the poller. Have you checked any firewalls or router ACLs along the path or changing from active to passive mode on the Agent? Lastly, have you checked the agent log files to see if there is any additional information available on the failure? 

0 Kudos

@aLTeReGo  one more query.. if Solarwinds is in HA then the agents should be configured to talk to VIP?

we tried manually and thru console too but no luck... this is case id in progress:00544107 

0 Kudos


@pratikmehta003 wrote:

@aLTeReGo  one more query.. if Solarwinds is in HA then the agents should be configured to talk to VIP?

we tried manually and thru console too but no luck... this is case id in progress:00544107 


Agents are fully HA aware and do not communicate through the VIP. In the event of a failover, any agents running in 'Active' mode will follow the active member of the pool. 

0 Kudos

@aLTeReGo  getting this error:

pratikmehta003_0-1592306449135.png

 

and how abt the other query i asked earlier for VIP showing under Agent settings page...

0 Kudos


@pratikmehta003 wrote:

@aLTeReGo  getting this error:

pratikmehta003_0-1592306449135.png

 

and how abt the other query i asked earlier for VIP showing under Agent settings page...


From the error in the screenshot, it appears as though the Agent port 17778 is not open, or that those services are not running on that server. Could that server IP address be the passive member? What happens when you attempt to access that same URL using IE on the server itself? 

0 Kudos

@aLTeReGo  not sure about the passive part.. it was working earlier and then suddenly stopped...

Its a Linux box so dont get a option to run via browser...

0 Kudos


@pratikmehta003 wrote:

@aLTeReGo  not sure about the passive part.. it was working earlier and then suddenly stopped...

Its a Linux box so dont get a option to run via browser...


I suggestion was to go to the Orion server itself that the Agent is registered with (or trying to register with), open IE, and go to that URL. Does it work? If so, then it sounds like there's something blocking the flow of traffic from the Linux machine to the Orion server. If it doesn't work, then it's likely that services aren't running on the Orion server at that URL. 

If you're not already, try using the IP address of the Orions server rather than a hostname or FQDN to eliminate possible name resolution issues. If you can, try pushing the agent from the Orion web console. 

0 Kudos

@aLTeReGo
Thanks for that info. But I also see that when we go to manage agents for manualy deploying agent, the IP address shown is thr VIP so that means we need to manually put the active member IP?
0 Kudos

@aLTeReGo yes i ran the init command and entered all information but says that Agent service start failed...port 17778 is open an no issues found..

Can u tell me the location for the agent logs?

0 Kudos

Agent Log files are located under /opt/SolarWinds/Agent/bin/appdata/Logs

0 Kudos

Hi @aLTeReGo 

I am seeing another strange thing.. when i try to deploy the agent, that device shows 100% packet loss and agent never gets deployed itself..

this is happening for 2 different customers.. any idea why this could be happening?

0 Kudos

If the node is appearing in Orion then it sounds like the Agent was installed and registered with the AMS. Is that node listed under 'Manage Agents'? 

0 Kudos

below snap from logs..

pratikmehta003_0-1590649783163.png

 

0 Kudos
Level 9

Love the idea, bit iffy on the reality....let me explain...

We install the rpm on the linux host and tie it to the poller. Then we monitor to view activity, and these are our findings:

When the agent is polled, it generates a flurry of activity that consumes nearly 100% of a single cpu. It is brief (about 2-3 seconds), then dies down. This has an impact on the 1min average, and sometimes on the 5min average uptime as well.

Furthermore, when using the agent, we constantly show 100% packet loss. Host is pingable, gathers stats, etc., but alas.

Hosts: RH 6x and 7x 64bit. SAM 6.9.1 and all other modules are latest (installed yesterday).

I ran stats for 2 hours on the system (it does nothing...absolutely nothing) as an snmp polled and another two hours as agent polled. SNMP reflected usage of system accurately (as well as 0% loss on packets), whereas agent constantly would report major spikes in cpu and uptime, though the only load on the system is/was the agent.

Note: during the statistics gathering in both situations, runq was mostly 0, and at times 1, but never any blocked processes.

Would love to know if there is something I'm doing wrong here, as it doesn't seem to shake to the surface very well.

0 Kudos

How does the Agent appear under Manage Agents? Is it in a healthy state? How is availability configured on the node? Are you using the Agent for availability calculations or ICMP? What happens when you change it? The behavior you're describing is very unusual, and perhaps a bit too complex to delve into on a forum such as Thwack. Have you opened a case with support regarding these issues?

0 Kudos

@aLTeReGo agent does appear but with error stating not successfull. I m using ICMP option only under List resources.

When i change the polling back to ICMP, there is no packet loss observed... Yes i have raised the case yesterday.. waiting for support to revert...

0 Kudos

So....after going back and forth with support for a week or so, we all stepped back and...oh yeah....the agent selects the first adapter in a list to ping, not the adapter that it is installed across. So, if your primary adapter is eth0 but shows up second, eth1 (nonrouted) will be the interface it attempts to ping, and will fail. But statistics for the server itself will continue to feed fine, as these are coming over your primary adapter.

So, now I know what to look for in the GUI to get it straightened out. Liking the agent so far--fixes the disk space issue with LInux real nice.

Level 16

aLTeReGo

One basic question when monitoring Linux device via agents...we get 2 options when we list resources right- one is ICMP-Fast Ping and other Status & Response time Agent...

I would like to know when will the device show as down if Status & Response time Agent option is selected...

0 Kudos

The two options function essentially identically. Agent status will report ‘down’ when a AMS message goes unreplied to by the Agent. The ICMP option reports a node being down when the node fails to reply to an ICMP echo request. The ‘faster’ in this case is really about payload size. ICMP packets are super tiny, where as AMS messages are full size HTTP requests which are a bit larger

Sent from my iPhone

0 Kudos