Just like in topic for future knowledge - Brocade CES/CER ethernet switches (L2+L3) can suffer from MPLS daemon restart because of LibreNMS Discovery in cron.d. I got my LDPs disconnected exactly every 6 hours. When “the time” was close I started to monitor device closely. Suddenly at 18:32 of my time the cpu reached almost 95% and caused process to do graceful restart (I saw this on Juniper’s logs). LibreNMS is the issue, still checking which module had the highest impact, but just letting you know if someone experiences similiar issue.