Polling partitioning / distributed polling
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
Hi Brad,
The web interface relies on local RRD files, regardless of the RRD Cache setup. The RRD Cache setup works only for the poller to push data to the master. If you want to use the web interface on the poller, you'll need to export the RRD folder on the master via NFS and mount it on the poller. We actually dedicate web nodes for the web interface that mount the NFS folder read-only. The pollers do not even get apache installed.
HTH,
Ahmed.
Le jeu. 5 nov. 2020, à 16 h 57, Brad Bendy via observium < observium@observium.org> a écrit :
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no
errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Hi Brad,
I'm not sure I understand the issue. You can add devices from any node running the UI (it all ends up in the database). If you wish for the device to be polled by a specific poller, you can assign it per device once declared. The only issue I had with this is that the 'web node' on which you access the UI needs to be able to ping the target device (or you check the 'skip ping' option).
HTH Ahmed.
Le ven. 6 nov. 2020, à 11 h 49, Brad Bendy via observium < observium@observium.org> a écrit :
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad
Bendy via observium
Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the
graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller
instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no
errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
The quickest and easiest is to add the device via the CLI on the poller you want to associate it with. That way you ensure it has the needed connectivity and that the poller is set in the DB without having to make changes via the UI.
Ron Culler
-----Original Message----- From: observium [mailto:observium-bounces@observium.org] On Behalf Of Brad Bendy via observium Sent: Friday, November 06, 2020 11:49 AM To: Observium observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Makes sense, easier to script as well for us. Ill just add it via CLI on the polling node.
Thanks for the help!
On Fri, Nov 6, 2020 at 10:03 AM Ron Culler Ron.Culler@securedesigns.com wrote:
The quickest and easiest is to add the device via the CLI on the poller you want to associate it with. That way you ensure it has the needed connectivity and that the poller is set in the DB without having to make changes via the UI.
Ron Culler
-----Original Message----- From: observium [mailto:observium-bounces@observium.org] On Behalf Of Brad Bendy via observium Sent: Friday, November 06, 2020 11:49 AM To: Observium observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
So doing more testing with this. When I add a device via CLI the poller never tries to poll it, looking at the DB it shows poller_id is 0. Is there something in the config on the poller server im missing? I don't see how it would get assigned tto then do the polling.
Looking at add_device.php I see no flags to pass to say "this is poller ID X". I see in the source there is a $config['poller_id'] so I set that to the number, but when adding a new device it does not show that number in the database.
Is there something I am missing or another set of docs maybe?
Thanks
On Fri, Nov 6, 2020 at 10:06 AM Brad Bendy brad.bendy@gmail.com wrote:
Makes sense, easier to script as well for us. Ill just add it via CLI on the polling node.
Thanks for the help!
On Fri, Nov 6, 2020 at 10:03 AM Ron Culler Ron.Culler@securedesigns.com wrote:
The quickest and easiest is to add the device via the CLI on the poller you want to associate it with. That way you ensure it has the needed connectivity and that the poller is set in the DB without having to make changes via the UI.
Ron Culler
-----Original Message----- From: observium [mailto:observium-bounces@observium.org] On Behalf Of Brad Bendy via observium Sent: Friday, November 06, 2020 11:49 AM To: Observium observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
poller_id is autogenerated by the partitioned/distributed poller installation based on a configured unique poller_name.
The documentation is available to enterprise users who email me asking for it :)
adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 12 November 2020 17:16 To: Observium observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing more testing with this. When I add a device via CLI the poller never tries to poll it, looking at the DB it shows poller_id is 0. Is there something in the config on the poller server im missing? I don't see how it would get assigned tto then do the polling.
Looking at add_device.php I see no flags to pass to say "this is poller ID X". I see in the source there is a $config['poller_id'] so I set that to the number, but when adding a new device it does not show that number in the database.
Is there something I am missing or another set of docs maybe?
Thanks
On Fri, Nov 6, 2020 at 10:06 AM Brad Bendy brad.bendy@gmail.com wrote:
Makes sense, easier to script as well for us. Ill just add it via CLI on the polling node.
Thanks for the help!
On Fri, Nov 6, 2020 at 10:03 AM Ron Culler Ron.Culler@securedesigns.com wrote:
The quickest and easiest is to add the device via the CLI on the poller you want to associate it with. That way you ensure it has the needed connectivity and that the poller is set in the DB without having to make changes via the UI.
Ron Culler
-----Original Message----- From: observium [mailto:observium-bounces@observium.org] On Behalf Of Brad Bendy via observium Sent: Friday, November 06, 2020 11:49 AM To: Observium observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So we just add via CLI on those nodes? I guess that's the best way?
Thanks!
On Fri, Nov 6, 2020 at 8:03 AM Adam Armstrong via observium observium@observium.org wrote:
Hi,
The poller instances aren't expected to have a web interface :)
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of Brad Bendy via observium Sent: 05 November 2020 21:57 To: Observium Network Observation System observium@observium.org Cc: Brad Bendy brad.bendy@gmail.com Subject: Re: [Observium] Polling partitioning / distributed polling
So doing some more digging, if I login to the master instance all the graphs are there. No graphs on the poller instance. I guess that makes sense as the poller instance has no access to the actual RRD files, is that then correct? Just only use the master instance for actually viewing anything?
Also just to make sure, we have to add the devices from the poller instance, only way to do that, correct?
Thanks
On Thu, Nov 5, 2020 at 11:47 AM Brad Bendy brad.bendy@gmail.com wrote:
Hi,
I've set this up per what I can find in the docs. Master server has RRDcached running, MySQL and web server. I have another server acting as poller server. I go to that server and add hosts, the host add find. All the graphs are failing and on the local server i t shows nothing in the RRD dir (which is right I would think), on the remote hosts it shows RRD files but when I do a debug I see "file already exists" and it does on the remote server.
Looking at the rrd.journal file I see updates being issued and see no errors.
On the polling server I have added: $config['rrdcached'] = "x.x.x.x:42217"; $config['rrd']['no_local'] = TRUE;
And updated cron to: */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 discovery --host new >> /dev/null 2>&1 # Run multithreaded poller wrapper every 5 minutes */5 * * * * observium /opt/observium/observium-wrapper -i 1 -n 0 poller >> /dev/null 2>&1
As I have one poller server and this is the first one.
I see no other errors in other logs but all graphs show "error drawing graph". Is there something else in my config I am missing to make this work?
Thanks
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
participants (4)
-
Adam Armstrong -
Ahmed Rahal -
Brad Bendy -
Ron Culler