We have observed looking glass servers becoming unresponsive after over 24 hours of continuous operation, sometimes 48 sometimes 72. The Axis Mundi backend appears to remain connected to the broker but Flask stops serving requests.
The only error observed so far is a Python malloc error suggesting a memory leak possibly in an underlying library or even within AM code itself.
We are creating a bug pending further investigation
We have observed looking glass servers becoming unresponsive after over 24 hours of continuous operation, sometimes 48 sometimes 72. The Axis Mundi backend appears to remain connected to the broker but Flask stops serving requests.
The only error observed so far is a Python malloc error suggesting a memory leak possibly in an underlying library or even within AM code itself.
We are creating a bug pending further investigation