Add back the cache to avoid generating spurious events when bootinfo
packets are lost; do not send the events if a node has requested bootinfo within the last 10 seconds; this implies that the bootinfo reply packet was lost and the node is retrying. This should quiet things a bit from stated, although eventually this should be dealt with in stated instead of with this hack. Mac originally added the cache to proxydhcp. I've added a cache to bootinfo to solve the same problem (proxydhcp is no longer used), although I have changed to implementation to use the DBM backend rather then a one-off hash table. dbopen() with a NULL file argument creates an in-memory database (no backing store), and I have used the DB_HASH db type. So, the hash table is done with a couple of calls to DBM routines.
Showing with 128 additions and 21 deletions