I have renamed my client’s but the old names and events continue to hang around on the uchiwa dashboard. I have restarted the sensu-server, uchiwa dashboard. I even restarted redis. Still the old events keep showing up.
I had to load redis-cli and flush all. Now its all neat. For some reason redis-server restart did not flushall.
···
On Friday, October 7, 2016 at 1:52:24 PM UTC+5:30, Siddharth Jagtiani wrote:
I have renamed my client’s but the old names and events continue to hang around on the uchiwa dashboard. I have restarted the sensu-server, uchiwa dashboard. I even restarted redis. Still the old events keep showing up.
While you can do this (the flush), it is kinda a heavy hammer.
Redis is persistent, restarting it will not flush it.
This is a consequence of Sensu’s design:
The sensu server doesn’t “know” that a check was renamed, so from its perspective the old event is still there, because it never got the memo (a resolve event) for it.
The same thing would happen if you renamed a host while it was red.
At Yelp use Puppet to deploy checks, and we have it notify an Exec that runs a “sensu-cli even resolve $check” whenever they are deleted. (via a rename or otherwise)
This keeps it tidy on a per-host basis. We also have another janitor that cleans up any event that hasn’t checks in for 2 weeks.
···
On Fri, Oct 7, 2016 at 1:34 AM, Siddharth Jagtiani jsiddharth@mcruiseon.com wrote:
I had to load redis-cli and flush all. Now its all neat. For some reason redis-server restart did not flushall.
On Friday, October 7, 2016 at 1:52:24 PM UTC+5:30, Siddharth Jagtiani wrote:
I have renamed my client’s but the old names and events continue to hang around on the uchiwa dashboard. I have restarted the sensu-server, uchiwa dashboard. I even restarted redis. Still the old events keep showing up.
On the dashboard the resolve too does not take it way. I think as of now flush is what works.
···
On Friday, October 7, 2016 at 8:09:16 PM UTC+5:30, Kyle Anderson wrote:
While you can do this (the flush), it is kinda a heavy hammer.
Redis is persistent, restarting it will not flush it.
This is a consequence of Sensu’s design:
The sensu server doesn’t “know” that a check was renamed, so from its perspective the old event is still there, because it never got the memo (a resolve event) for it.
The same thing would happen if you renamed a host while it was red.
At Yelp use Puppet to deploy checks, and we have it notify an Exec that runs a “sensu-cli even resolve $check” whenever they are deleted. (via a rename or otherwise)
This keeps it tidy on a per-host basis. We also have another janitor that cleans up any event that hasn’t checks in for 2 weeks.
On Fri, Oct 7, 2016 at 1:34 AM, Siddharth Jagtiani jsidd...@mcruiseon.com wrote:
I had to load redis-cli and flush all. Now its all neat. For some reason redis-server restart did not flushall.
On Friday, October 7, 2016 at 1:52:24 PM UTC+5:30, Siddharth Jagtiani wrote:
I have renamed my client’s but the old names and events continue to hang around on the uchiwa dashboard. I have restarted the sensu-server, uchiwa dashboard. I even restarted redis. Still the old events keep showing up.