On 07/16/2008 10:21 AM, Mark Andrews wrote: >> Hi all, >> >> after upgrading to bind-9.4.3b2 I see the following in the logfile >> >> 16-Jul-2008 08:55:11.547 general: sockmgr 25af70: maximum number of FD >> events (64) received >> 16-Jul-2008 08:57:11.796 general: sockmgr 25af70: maximum number of FD >> events (64) received >> 16-Jul-2008 08:57:11.797 general: sockmgr 25af70: maximum number of FD >> events (64) received >> 16-Jul-2008 09:37:17.581 general: sockmgr 25af70: maximum number of FD >> events (64) received >> 16-Jul-2008 09:37:17.677 general: sockmgr 25af70: maximum number of FD >> events (64) received >> >> % uname -a >> SunOS mantis 5.9 Generic_122300-02 sun4u sparc SUNW,Sun-Fire-V210 >> >> In lib/isc/unix/socket.c I see a comment in #ifdef USE_KQUEUE >> * This is not an error, but something unexpected. If this >> * happens, it may indicate the need for increasing >> * ISC_SOCKET_MAXEVENTS. >> >> but there is no such comment in #elif defined(USE_DEVPOLL) >> >> Since this is a Solaris 9 system, I assume bind uses /dev/poll .. >> Is it something I should worry about? > > It just makes named slightly less efficient. > >> Do I need to increase ISC_SOCKET_MAXEVENTS? > > You could. >
I increased ISC_SOCKET_MAXEVENTS to 128 and I now see these warnings a few times a day (in contrast with many times in an hour, when it was 64). Sot.
