Behaviour of named with max-journal-size

[Anthony Moore]

Hi,

I have an instance of named (9.9.3-P1-ESV) with a 150M zone (unsigned) and max-journal-size 100M with frequent dynamic updates (about 1 every second). 

Occasionally CPU usage for the named process will spike up to maximum and named becomes slow to respond to queries; some even time out. This period lasts for about 10 to 15 minutes. lsof -p shows lots (varying between 5 to 15) of open file handles to the zone journal file, all at the same offset (possibly the same file descriptor shared between all the threads?). Once this period passes all the file descriptors are released and behaviour/performance of named goes back to normal.

Does anyone have any ideas what might be going on? 

Many thanks,

Anthony Moore
ajm at axdf.net
--