Why are Gluster bricks failing to start with the error 'Too many opened files'?

Issue

• In the output of the gluster v status commands, a bunch of bricks are down. Their status is N/A:

  Status of volume: vol_ba3db22a379d1cb58515337e1eedd4e3
  Gluster process                             TCP Port  RDMA Port  Online  Pid
  ------------------------------------------------------------------------------
  Brick 192.168.1.1:/var/lib/heketi/mounts/vg_
  39ccc70fabf2e144a5b0909eacffef29/brick_099f
  fced2cd326324331e97152da8c40/brick          N/A       N/A        N       N/A
  Brick 192.168.1.2:/var/lib/heketi/mounts/vg_7
  40c54c786f55628a7b177dec970c5fe/brick_a9f05
  76d88a4749ac73d78a4cf0fbeb5/brick           49157     0          Y       550
  Brick 192.168.1.3:/var/lib/heketi/mounts/vg_6
  3eca5a563dbd4f735d93fac0766d0f2/brick_fd83f
  251e012be0251b17ea8c9938b6a/brick           N/A       N/A        N       N/A
  Self-heal Daemon on localhost               N/A       N/A        Y       2762
  Self-heal Daemon on 192.168.1.3             N/A       N/A        Y       2144
  Self-heal Daemon on 192.168.1.1             N/A       N/A        Y       2795
  

• Executing the command gluster v start VOLNAME force doesn't help. Some bricks are recovered, but some other bricks go down instead.

• Reviewing the brick logs, available at /var/log/glusterfs/bricks, there are lots of Too many opened files error messages:

   W [MSGID: 113075] [posix-helpers.c:2113:posix_fs_health_check] 0-vol_ba3db22a379d1cb58515337e1eedd4e3-posix: open_for_write() on /var/lib/heketi/mounts/vg_63eca5a563dbd4f735d93fac0766d0f2/brick_fd83f251e012be0251b17ea8c9938b6a/brick/.glusterfs/health_check returned ret is -1 error is Too many open files [Too many open files]
  

• Why is this issue occurring? How to solve this problem?