This happens now and then without anything in the logs or other clear reason.
Whats happening
Suddenly a domain can stop resolv, the dns-servers are synced, all of them have the data and the zone-files looks good.
Restarting the servers, services, bind, making sure everything is updated and so on does not change anything.
Only solution is to remove the domain from DNS, and add it again - and sync the cluster.
For some time its the same domain doing it, after adding more domains to the system the issue moves to another domain.
Anyone else seen this?
I manage 8 myvesta setups and they all have this issue, not at the same time but very random. Sometimes its a few weeks between the issue.
Anything i should look for next time it happens? log-files or something that might help finding the issue?
Zone-file for a domain stops working (SRV records format?)
- myVesta
- Site Admin
- Posts: 948
- Joined: Fri Jun 19, 2020 9:59 am
- Has thanked: 9 times
- Been thanked: 6 times
Restart bind
and see what error will come up in /var/log/syslog
Probably you have some record with an invalid type of value.
For example, you entered a hostname in the 'A' type of the record or something like that.
Code: Select all
systemctl restart bind
Probably you have some record with an invalid type of value.
For example, you entered a hostname in the 'A' type of the record or something like that.
-
- Posts: 12
- Joined: Sun Mar 26, 2023 8:52 pm
Redacted this post, it was just some logfiles and stuff before i found the issue - see next post.
Last edited by NordicSnowman on Fri Jun 23, 2023 5:17 am, edited 1 time in total.
-
- Posts: 12
- Joined: Sun Mar 26, 2023 8:52 pm
After some reading on the SRV record, i think this is a bug in MyVesta ? (and vesta too, i remember having this since long time ago)
In my first post i found out that all domains with issues had a SRV-record.
I tried adding a test domain, named testdomain.com
Everything looked nice.
Then i added this to a SRV record
pasted exactly like that.
But in the zone-file, its saved like this
So i reloaded bind and looked in the syslog
If i edit the SRV-post, and remove the ", and save it.
Now the row looks like this in the zonefile db-file
The bind-server reloads perfectly again.
Whats probably happening is that when i add this to a domain, it works until the serial/timestamp of the zonefile expires, and then the domain dies.
Probably a bug, Vesta/MyVesta (and probably hestia too) adding " to the data field on SRV-records.
In my first post i found out that all domains with issues had a SRV-record.
I tried adding a test domain, named testdomain.com
Everything looked nice.
Then i added this to a SRV record
Code: Select all
20 5000 sip-server.example.com.
But in the zone-file, its saved like this
Code: Select all
srv-test 14400 IN SRV 10 "20 5000 sip-server.example.com."
Code: Select all
Jun 23 05:13:07 panel named[239845]: dns_rdata_fromtext: /home/admin/conf/dns/testdomain.com.db:15: not a valid number
Jun 23 05:13:07 panel named[239845]: zone testdomain.com/IN: loading from master file /home/admin/conf/dns/testdomain.com.db failed: not a valid number
If i edit the SRV-post, and remove the ", and save it.
Now the row looks like this in the zonefile db-file
Code: Select all
dreamhost._domainkey 14400 IN SRV 10 20 5000 sip-server.example.com.
The bind-server reloads perfectly again.
Whats probably happening is that when i add this to a domain, it works until the serial/timestamp of the zonefile expires, and then the domain dies.
Probably a bug, Vesta/MyVesta (and probably hestia too) adding " to the data field on SRV-records.
- myVesta
- Site Admin
- Posts: 948
- Joined: Fri Jun 19, 2020 9:59 am
- Has thanked: 9 times
- Been thanked: 6 times
Hmmm, something other is the problem there because I successfully used SRV records many times.
Actually, I always followed this format:
I think the catch is putting all three numbers in the "Priority" field.
Actually, I always followed this format:
I think the catch is putting all three numbers in the "Priority" field.
-
- Posts: 12
- Joined: Sun Mar 26, 2023 8:52 pm
Nice, that might explain the issue
Edit: changed now on all five servers with issues, and they all report normal function in syslog on bind.
Thank you for correcting my stupidity, i should have known
Edit: changed now on all five servers with issues, and they all report normal function in syslog on bind.
Thank you for correcting my stupidity, i should have known