Bind 9.9.2 ADB Question Update

Cathy Almond cathya at isc.org
Thu Nov 15 16:43:47 UTC 2012


On 15/11/12 16:17, Cathy Almond wrote:
> On 15/11/12 15:49, Manson, John wrote:
>> The adb grow-names process? does not appear to be related to recursive cache as I cleared cache while monitoring syslog and the counter kept increasing.
>> However a reload did start the adb grow-names process anew.
>> Both shown below
>>
>> .
>> .
>> .
>> Nov 14 15:25:40 local at mercury named[2920]: [ID 873579 daemon.notice] all zones loaded
>> Nov 14 15:25:40 local at mercury named[2920]: [ID 873579 daemon.notice] running
>> Nov 14 15:27:15 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 1531 starting
>> Nov 14 15:27:15 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>> Nov 14 15:28:40 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 2039 starting
>> Nov 14 15:28:40 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>> Nov 14 15:30:27 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 3067 starting
>> Nov 14 15:30:27 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>> Nov 14 15:32:38 local at mercury named[2920]: [ID 873579 daemon.info] received control channel command 'flush'
>> Nov 14 15:32:38 local at mercury named[2920]: [ID 873579 daemon.info] flushing caches in all views succeeded
>> Nov 14 15:40:43 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 4093 starting
>> Nov 14 15:40:43 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>> Nov 14 15:46:41 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 6143 starting
>> Nov 14 15:46:41 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>> Nov 14 16:01:11 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names to 8191 starting
>> Nov 14 16:01:12 local at mercury named[2920]: [ID 873579 daemon.info] adb: grow_names finished
>>
> 
> Hope this helps:
> 
> https://kb.isc.org/article/AA-00548/30/Why-is-BIND-logging-adb%3A-grow_entries-to-and-adb%3A-grow_names-to-.html

KB article updated to note that when cache (including ADB) is flushed
via rndc flush, the hash tables are not made smaller again.  (There is
no reason to expect that the cache will not need the larger sizes again
if it needed them before).

So in John's example above, the cache grew for ~7 minutes before it was
flushed.  It started repopulating again after the flush, and ~8 minutes
later had reached and exceeded its earlier levels, and the hash tables
were once again increased.

Cathy



More information about the bind-users mailing list