Strange problem : - entering new DNS zone using ispconfig. - DNS_SOA table has now 207 records (newly added zone is added) - file named.conf.local contains all active records EXCEPT last two - on first try file err.pri.zonename... is created - after editing DNS record once again correct pri.zonename.extension file is created on both DNS servers Of course DNS doesn't resolve last two added zones. I compared everything that came into my mind, but cannot find solution. ISPConfig is updated to the last version. Any suggestion is appreciated.
Already found SELECT origin, xfer, also_notify, update_acl, dnssec_wanted FROM dns_soa WHERE active = 'Y' in bind.plugin.inc.php but this SQL is returning all records, including two missing ones.
Use the debug mode to get the exact messages when the zone gets created and use named-checkzone command to find out why BID refused to load the zone (which results in the .err file). A first refusal is ok, but after the last record, the zone should be accepted by bind, unless a record is missing.
@Th0m : sorry if I wasn't clear enough. What I meat is that i found the function, which creates named.conf.local file. In this function there is select SQL, which I ran manually and it shows all records, including last two missing. @till : Two days ago I manually edited named.conf.local file in DNS servers and domain was successfully registered afterwards. Today I edited DNS record again and affected domain is missing again (actually two). Before that I put ispconfig in debug mode, but unfortunately there's nothing usable in log file.
I enabled debug logs, but unfortunately this doesn't enable logging on DNS server. Is there some manual way to do it ?
I opened MySQL command prompt on DNS server and listed records in table DNS_SOA. At my big surprise records in this table and on main server are not the same - and of course a record for affected zone is missing. On both DNS servers... How to diagnose where problems occured and how to synchronize records from "master" table ? There are 3 virtual servers, one mail/web/db and 2 dns servers. Resync from ispconfig unfortunately doesn't do the job.
named.conf.local may not be edited manually, so any update in any DNS must remove the manual changes you made as you did not fixed the underlying issue. Enable debug log level for this dns slave server on the master server and then run server.sh on the slave to find out why the slave can't sync DNS records anymore. https://www.faqforge.com/linux/debugging-ispconfig-3-server-actions-in-case-of-a-failure/
Found the culprit... It seems like last ispconfig update added a field "dnssec_algo" into table dns_soa. This field was added to master, but not to DNS databases... I added field to the table manually and forced resync, now everything working. But somehow I failed to perform ispconfig upgrade on DNS servers, saying that root password for master database is incorrect, although 100% entering correct one. But this issue is not connected to ispconfig itself. I tried manually running "server.sh", but all I got was "finished ok" Anyway, everything works fine, now must only find a way to update ispconfig on DNS servers.
Yes, master and all slave servers must always use the same ISPConfig version, which means you must update all systems, not just the master. Otherwise replication between the systems will fail.
