+ Implementing/Enabling diskquota suspected to cause mirror failures and after 2 subsequent mirror failures diskquota is disabled.
+ Mirror failures might not showup in logs on Mirror or Primary segment pairs but seen in the gp_segment_configuration & gp_configuration_history tables to confirm mirrors failed and are down.
Steps lead to reported issue:
1.] gpconfig -c shared_preload_libraries -v 'diskquota-2.2'
2.] gpstop -ar
3.] then cluster went to restart and in the middle of the restart process it prompted for choosing (i) immediate (or) (f) fast mode.
4.] "fast(f)" chosen and then cluster restarted making one of the mirror segment instance go into a failed state.
Greenplum 6.24.6
Diskquota 2.2
No RCA concluded from available logs.
Instead of issuing a database restart with gpstop -ar database was stopped with `gpstop -a` and then started with `gpstart -a` which successfully enabled the diskquota without any failures.