Urgent: Mad Bind

Jim Reid jim at rfc1035.com
Fri Jan 28 17:15:10 UTC 2000


>>>>> "Birgit" == Birgit Schmid <Birgit.Schmid at ecrc.de> writes:

    >>  Could it be that you've just caught the name server while it
    >> was in the middle of cleaning up its cache? That could account
    >> for a lot of user-mode CPU activity for a few seconds every
    >> hour or so.

    Birgit> is there anything to state this theory?

Well the name server has a cache which it preens regularly to get rid
of expired resource records. If you have a big cache and a slow CPU,
cleaning up the cache might take noticeable amounts of CPU time. And
since your system's statistics indicated CPU was executing in user
mode, it suggests the name server was busy doing something that didn't
need it to make many system calls. Like handling large numbers of
queries for instance.

    Birgit> I found some files named xfer.ddt.*. after deleting them
    Birgit> and a restart bind feels a lot better.

    Birgit> Can anyone enlighten me on this?

Beats me. Maybe there was something wrong with the zone transfers that
created these xfer.ddt. files.



More information about the bind-users mailing list