Am new to graphite monitoring tool. I have one question in this setup. I have two servers and hear one server treated as a hosted server(installed graphite,collectd,statsd and grafana)and it grafana displays the all metrics. In the another second server i have installed the graphite and collectd.Now i would need to send this second server collectd informtion to first server(hosted server)and those metrics information will need to display the web using grafana...
could you please suggest me is there any plugin or any way to setup this configuration?
Thanks.
You don't actually need graphite on the second host, you can just configure collectd on that host to write to graphite (actually the carbon ingest api) on the first host.
https://collectd.org/documentation/manpages/collectd.conf.5.shtml#plugin_write_graphite
If you do want to have graphite on both servers for some reason, you can use multiple Node entries in your collectd config to have it send metrics to both graphite instances.
Related
I have a Google Kubernetes Engine running in the Google Cloud Platform.
I would now like to know how many HTTP requests are received by our Kubernetes cluster. I'd like that to be displayed in Google Stackdriver.
Unfortunately I can't find any appropriate metric in the documentation at https://cloud.google.com/monitoring/api/metrics_kubernetes.
Is there a way to get a chart with the count of all incoming HTTP request to a GKE cluster in Stackdriver?
Unfortunately, that's a limitation of Kubernetes itself - it doesn't expose very much information about the services running in it. You'll need to either install Istio and configure the Stackdriver adapter (see my post on this) or use something like OpenCensus in your app to create a custom metric. Another option would be to create a log-based metric to count the requests.
I want to set up a server (hosted on aws/or a running system in some part of the world) as an NTP server that can be queried globally.
Currently, I have modified the ntp.conf file on the node to be made the server as server . But the problem is, on using an NTP client if I try to query time from this server, or rather on using sudo ntpdate it says no suitable server found.
However, if I replicate the same on my local network (the server, as well as the querying node, are all on the local network) then this works perfectly fine.
I think the problem might lie in the ntp.conf file. Do I need to put some specific restrict lines for this to work publicly as well? And no I cannot list the server on public ntp pages. Is it at all possible?
Solved. This was a port issue. I was testing it on aws and had to manually open the related udp ports.
What is the best way to send metrics from multiple servers to graphite (will then use Grafana to view graphite data):
1) Install collectd daemon on all servers and send to a collectd central server. Configure collectd server to send data to graphite.
2) Install collectd daemon on each all servers and configure to send metrics directly to graphite server.
Thank you
I have a graphite relay and webapp installed on one servers, that is supposed to be communicating with 4 carbon caches (and respective webapps) on 4 other servers. I've validated that the relay is working by observing that different whisper files are being updated on different carbon-relay servers.
However, the webapp is only showing metrics that are stored on the first carbon cache server in the list and I'm not sure what else to look at.
The webapps on the carbon relays are set up to listen on port 81, and I have the following in local_settings.py on the relay server (the one I'm pointing my browser at):
CLUSTER_SERVERS = ["graphite-storage1.mydomain.com:81", "graphite-storage2.mydomain.com:81", "graphite-storage3.mydomain.com:81", "graphite-storage4.mydomain.com:81", ]
However - at one point I did have all metrics on all servers - I've migrated from a single instance to this federated cluster. I've since removed the whisper files that weren't active on each carbon-cache server. I've restarted all carbon-caches, the carbon-relay and the webapp server several times. Is there somewhere the metrics-->carbon-cache mapping is getting cached? Have I missed a setting somewhere?
Using BizTalk I need to read data from one of two databases that are hosted in Unix using ODBC.
The data is replicated between the databases and if one of the databases does not respond I need to switch to the other. There is no load balancer or anything so I need to be able to do the switch on the BizTalk server.
I was thinking of creating two receive locations, one for each database server, only one of them enabled and then have a Windows service that periodically tries to make a connection to one of the database servers and if there is an exception, call a powershell script that disables the receive location for the server that does not respond and enable the other receivelocation.
Is there a better solution for this?
I would solve this as follows:
In Biztalk create a single http receive location.
Create a windows service
In the windows service poll the first database, if it does not respond poll from the second database
Have the Biztalk service post the information to the http receive location
You need to consider what happens if you read the same data twice, once from the main database and once from the backup.