Scenario is that large number of PCEP connections are established between Controller and PCCs (approx. 50) and large number of LSPs are delegated by each PCC (approx. 500). Then PCEP connections are brought down and back up, and this process repeated in quick succession, by restarting PCCs (pcc-mock used for repro). This results in PCEP sessions management and stats collection threads running into deadlock. There are 3 different deadlock scenarios seen. Based on feedback of previous combined patch, I am labelling them instance 1, 2 and 3 and creating separate patches for each (details in comments section below).