Hi, we have two dns server runninng bind 8 with a lot of zones (near 640000) and we finally decide to migrate to pdns + bind backend. All seems really fast and service comes up in 4-5 minutes vs. 2h40m with bind8 with the same zones and named.conf. But suddenly (from 10m to 1h30m not more) and not in a specific situation, process dies with this trace:
Oct 27 14:24:31 dns pdns[29788]: Got a signal 11, attempting to print trace: Oct 27 14:24:31 dns pdns[29788]: /usr/sbin/pdns_server-instance [0x4771b2] Oct 27 14:24:31 dns pdns[29788]: /lib/libc.so.6 [0x2b7ad835e110] Oct 27 14:24:31 dns pdns[29788]: /usr/lib/libstdc++.so.6(_ZNSs6assignERKSs+0x2e) [0x2b7ad7e2676e] Oct 27 14:24:31 dns pdns[29788]: /usr/sbin/pdns_server-instance(_ZN12Bind2Backend11queueReloadEP13BB2DomainIn fo+0x131) [0x4ba071] Oct 27 14:24:31 dns pdns[29788]: /usr/sbin/pdns_server-instance(_ZN12Bind2Backend18DLReloadNowHandlerERKSt6ve ctorISsSaISsEEi+0x25d) [0x4bae2d] Oct 27 14:24:31 dns pdns[29788]: /usr/sbin/pdns_server-instance(_ZN11DynListener11theListenerEv+0x41b) [0x486fbb] Oct 27 14:24:31 dns pdns[29788]: /usr/sbin/pdns_server-instance(_ZN11DynListener17theListenerHelperEPv+0x9) [0x487c19] Oct 27 14:24:31 dns pdns[29788]: /lib/libpthread.so.0 [0x2b7ad821ff1a] Oct 27 14:24:31 dns pdns[29788]: /lib/libc.so.6(__clone+0x72) [0x2b7ad83f85d2] We also compile the pdns-2.9.21-svn.20061001.897 but the crash is the same. This situation is similar in redhat (centos 4.4 x86_64) or debian (etch x86_64)distribution in the same machine (dell 1950 with 8 giga ram), whith compiled o rpm version. For a try I convert, with zone2sql, to gmysql and it seems to run stable, but, for now, we cannot use this backend because the provisioning software wich uses old bind zone files. Regards, = = = = = = = = = = = = = = = = = = = = = = = Marco Chiavacci = = = = = = = = = = = = = = = = = = = = = = = _______________________________________________ Pdns-users mailing list [email protected] http://mailman.powerdns.com/mailman/listinfo/pdns-users
