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
We recently updated from Cacti 0.8.8 to 1.2.24 and also updated the Weatermap plugin. Unfortunately, the performance of the new Cacti on the same hardware was suddenly very poor. The poller was constantly killed after 59 seconds because of max execution time. As soon as the plugin was deactivated, the poller was finished after 7 seconds.
We looked at the plugin code afterwards and found that the foreach in 1 takes forever. If we understand it correctly, this part creates error logs for every broken graph. Unfortunately, we have a few of these. The foreach is executed no matter what the log level in the plugin configuration is (quiet/chatty).
As soon as we commented out these 4 lines, the poller was finished in 43 seconds and all our weathermaps worked again.
As mentioned above, I'm not sure if this is really a bug. After all, the broken graphs are the main cause. But since it took us some time to figure this out, it might be helpful for others. Also, whether this log foreach should be run or not when the weatermap log level is quiet (not cacti log level), I am a bit unsure.
The text was updated successfully, but these errors were encountered:
That should only be at the debug level. Is that how you are running your Cacti? There will of course be issues running at Debug, especially if your storage is constrained and you have a lot of poller items.
Maybe the request is to have a separate logging function for WeatherMap like we have for boost so that you don't wreck your server if you are debugging other things. What do you think?
I am not sure if this is really a bug.
We recently updated from Cacti 0.8.8 to 1.2.24 and also updated the Weatermap plugin. Unfortunately, the performance of the new Cacti on the same hardware was suddenly very poor. The poller was constantly killed after 59 seconds because of max execution time. As soon as the plugin was deactivated, the poller was finished after 7 seconds.
We looked at the plugin code afterwards and found that the foreach in 1 takes forever. If we understand it correctly, this part creates error logs for every broken graph. Unfortunately, we have a few of these. The foreach is executed no matter what the log level in the plugin configuration is (quiet/chatty).
As soon as we commented out these 4 lines, the poller was finished in 43 seconds and all our weathermaps worked again.
As mentioned above, I'm not sure if this is really a bug. After all, the broken graphs are the main cause. But since it took us some time to figure this out, it might be helpful for others. Also, whether this log foreach should be run or not when the weatermap log level is quiet (not cacti log level), I am a bit unsure.
The text was updated successfully, but these errors were encountered: