Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

Proxysql Cluster mode generating too much logs

37 views
Skip to first unread message

Peter Lyoo

unread,
Jan 23, 2025, 3:18:41 PMJan 23
to proxysql
Hi there,

I'm running ProxySQL in a Kubernetes environment and recently switched to cluster mode. However, it is generating excessive logs about the cluster connection status, such as [INFO] Cluster ~~~. These logs are increasing exponentially with the number of pods, which is impacting our log collection pipeline.

Is there a way to suppress these INFO-level logs, or at least prevent the cluster status logs from being printed to stdout?

Thanks,
Peter

René Cannaò

unread,
Feb 3, 2025, 5:35:36 PMFeb 3
to Peter Lyoo, proxysql
Hi Peter,

I don't know what these log entries are pointing to, but I would recommend reading and analyzing the reason behind them, instead of just suppressing them.
Thus it would be more useful to share a large snippet of error log.

Thanks,
René

--
You received this message because you are subscribed to the Google Groups "proxysql" group.
To unsubscribe from this group and stop receiving emails from it, send an email to proxysql+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/proxysql/26cd4804-7207-4cb1-a54c-1457fe9e7b56n%40googlegroups.com.

Peter Lyoo

unread,
Mar 12, 2025, 6:28:02 AMMar 12
to René Cannaò, proxysql

Hi René,

Apologies for the delayed response. I missed your reply.

The log entries I was referring to are ProxySQL cluster connection success logs.
 It seems that INFO-level logs are generated for every new link established between nodes.

For example:
| 2025-03-12 10:11:36 [INFO] Received PROXYSQL CLUSTER_NODE_UUID from x.xx.xxx.xxx:42198 : xxxxxxxxxx4e7d-97c6-5f191dca1c71 0.0.0.0:6032 |
| 2025-03-12 10:11:36 [INFO] Created new link with Cluster node x.xx.xxx.xxx:42198 : xxxxxxxxxx4e7d-97c6-5f191dca1c71 at interface 0.0.0.0:6032 |

Since our system frequently scales in and out, these logs, along with other INFO-level messages, generate significant noise.
While INFO logs are useful, having the ability to control verbosity would be beneficial, especially given the impact on our log collection pipeline and costs.

To address this, I explored existing options and came across this PR, which introduces log verbosity controls.
Disabling INFO-level logs in some environments has reduced log collection costs by approximately 30%.

Would you consider adding an option to adjust log verbosity to help mitigate this issue?

Thanks,
Peter

Reply all
Reply to author
Forward
0 new messages