On Fri, Jan 25, 2008 at 02:48:08PM +0100, Claudio Jeker wrote: > On Fri, Jan 25, 2008 at 02:06:49PM +0100, Erich Hohermuth wrote: > > Hi > > > > > ok, but why is there no answer? > > > > Cause cctld.tix.ch which is one of the .ch nameservers doesn't give the > > right answer. > > > > > > ;bilder-floriannussbaumer.ch. IN NS > > > > ;; AUTHORITY SECTION: > > ch. 3600 IN SOA domreg.nic.ch. > > helpdesk.nic.ch. 2008012300 3600 1800 2592000 3600 > > > > I think you have to ask Switch how long it takes for the next reload > > cycle. > > > > This is fixed now. For some reasons nsd no longer updated the zone > internaly after 40 days of uptime. >
For the curious ones: It seems that nsd grows bigger and bigger over time if the zone is updated via IXFR instead of being reloaded once in a while. So after 40 days it hit the process memory limit and the zone transfers failed. I modified the cronjob to patch the local domain db from using nsd-patch to nsdc patch -- the latter calls nsd-patch and reloads the db afterwards. Here some syslog porn: Jan 25 13:05:22 ccTLD nsd[70689]: signal received, reloading... Jan 25 13:05:22 ccTLD nsd[86414]: memory recyclebin holds 73492 bytes Jan 25 13:05:22 ccTLD nsd[86414]: malloc failed: Cannot allocate memory Jan 25 13:05:22 ccTLD nsd[70689]: handle_reload_cmd: reload closed cmd channel Jan 25 13:05:22 ccTLD nsd[70689]: Reload process 86414 failed with status 256, continuing with old database Jan 25 13:05:22 ccTLD nsd[13262]: xfrd: zone ch.: soa serial 2008012501 update failed restarting transfer (notified zone) -- :wq Claudio _______________________________________________ swinog mailing list [email protected] http://lists.swinog.ch/cgi-bin/mailman/listinfo/swinog
