You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am working on a project that will run an agent in a sidecar in a ProxySQL pod in our kubernetes cluster, and had a couple small feature requests.
PAUSE status
We use PROXYSQL PAUSE (among some other commands) to gracefully stop our pods and drain traffic from them, and it would be super helpful if the status of this was returned via a sql command in the admin interface. Something like SELECT paused FROM proxysql_servers (kind of just making stuff up here, I'd leave it to the experts to determine the best course here). It'd help greatly with our health probes if we could expose that to the agent.
I could attempt to connect to ProxySQL on port 6603 as part of the probe, but we have so many backends that I'd have to cycle through them all to ensure that the system is paused (as opposed to a backend being shunned) and it seems like a hacky solution.
Replicated custom settings
We are using ProxySQL in clustering mode, with 3 core pods and 18+ satellite pods; the core pods push configs out to the satellite pods, and the satellite pods do not push changes back up to the core pods.
I was thinking it'd be handy to have some kind of custom variables in the database that get propagated to the satellite pods; I attempted to set admin-custom_flag in the global variables table on a core pod, but when I ran LOAD ADMIN VARIABLES TO RUNTIME it did not get sent out to the cluster.
I know that's a bit outside of the forte of ProxySQL, but it seems like custom variables that ProxySQL completely ignores but other things could use should be fairly low impact from a risk standpoint. And it'd prevent me from having to setup a consul cluster or something like it.
The text was updated successfully, but these errors were encountered:
New variable mysql_listener_paused added to table stats_mysql_global .
The variable is a boolean:
- true : listener is paused because PROXYSQL PAUSE was executed
- false : listener is not paused
The variable is also reported by Prometheus exporter as proxysql_mysql_listener_paused
Related to #4391
Hello!
I am working on a project that will run an agent in a sidecar in a ProxySQL pod in our kubernetes cluster, and had a couple small feature requests.
PAUSE status
We use
PROXYSQL PAUSE
(among some other commands) to gracefully stop our pods and drain traffic from them, and it would be super helpful if the status of this was returned via a sql command in the admin interface. Something likeSELECT paused FROM proxysql_servers
(kind of just making stuff up here, I'd leave it to the experts to determine the best course here). It'd help greatly with our health probes if we could expose that to the agent.I could attempt to connect to ProxySQL on port 6603 as part of the probe, but we have so many backends that I'd have to cycle through them all to ensure that the system is paused (as opposed to a backend being shunned) and it seems like a hacky solution.
Replicated custom settings
We are using ProxySQL in clustering mode, with 3 core pods and 18+ satellite pods; the core pods push configs out to the satellite pods, and the satellite pods do not push changes back up to the core pods.
I was thinking it'd be handy to have some kind of custom variables in the database that get propagated to the satellite pods; I attempted to set
admin-custom_flag
in the global variables table on a core pod, but when I ranLOAD ADMIN VARIABLES TO RUNTIME
it did not get sent out to the cluster.I know that's a bit outside of the forte of ProxySQL, but it seems like custom variables that ProxySQL completely ignores but other things could use should be fairly low impact from a risk standpoint. And it'd prevent me from having to setup a consul cluster or something like it.
The text was updated successfully, but these errors were encountered: