In a large cluster (24 nodes), the components have very often been timeout in their imm api calls. This timeout results in a component restart. During single step upgrade, due to the imm timeout, the amf locking nodegroup operation fails because of many component restarts escalated to node failover, etc... This ticket suggests IMMSV_DEFAULT_FEVS_MAX_PENDING and IMMSV_WAIT_TIME to be environment variables so the components can adjust these parameters according to the size of cluster.
commit eaf187e9a04ee147dacee8d722c09aecf9a52426
Author: Minh Chau minh.chau@dektech.com.au
Date: Tue May 18 14:41:59 2021 +1000
commit fa8325e924060496184cb57e43bd7eff3bb71a84
Author: Minh Chau minh.chau@dektech.com.au
Date: Tue May 18 14:41:59 2021 +1000