First off, I’d like to say I’ve been setting up my first VPS for the past few weeks and Virtualmin has been a huge help. Once I finally stopped using it as the Webmin module and upgraded the OS so I could install it cleanly with the script, it has been a breeze getting everything going.
One problem I’ve been having though, is that every day at seemingly random times, the BIND DNS server stops and all the websites become unreachable until I manually start it up again!
If anyone has had this problem and solved it, or if you have an inkling as to why it’s happening, that would be a huge help. Also, let me know what information/logs I can post that will assist in tracking down this problem. I’ll post some info and logs below.
Again, thanks to the developers (and community) for creating such an excellent product and providing a free version for public use.
I’ve unfortunately never heard of anything like that… the BIND DNS server is generally pretty reliable.
However, where I’d start is probably by looking in the log files. Which ones specifically depends on what distro you’re using, but I’d go into /var/log, and around the time that the DNS server stopped working, see if any of the logs in there show any errors of some kind that may explain what’s happening.
Sorry, I got interrupted before I got a chance to post more info regarding the server.
I’m running on Ubuntu 10.01. The DNS server has gone down at approximately the following times (I only get notified every 30mins, so it would have gone down within 30mins of the times):
24th Oct, 4:11pm
23rd Oct, 10:23am
21st Oct, 8:16am
20th Oct, 6:31pm
20th Oct, 5:31am
etc.
I looked through all of the log files in /var/log but found no errors that seemed related to this around those times. The only error I could find in /var/log/syslog is the following one (repeated many times):
Yeah, I’m not really seeing anything there that would explain the symptoms you’re describing. The “network unreachable” errors are really unusual, but if everything else is working, that doesn’t sound like the problem.
Is the BIND daemon actually not running? Or is it running, but not working properly?
I googled the “network unreachable” error and apparently that has to do with the server attempting to use IPv6 transport even though it doesn’t have IPv6 connectivity, or something along those lines, but that doesn’t really help me with the BIND server going down.
The BIND daemon goes down when the BIND server goes down, ie. at seemingly random times.
Also, I’m not 100% sure but I think in Virtualmin by default BIND logs info messages and above in the syslog. At least that’s what it looks like when I go to Servers -> BIND -> Logging and Errors. I tried setting a new log to /var/log/bind at debug level (do I have to enter a number in here, if so which is the number for the most info to be logged?) and selected yes for all options, but hasn’t logged any info there yet.
There are some issues with having log BIND to files under /var/log. First, the directory must have the correct permissions and ownership set. Then, if your distro uses “AppArmor”, which is a tool to limit system access rights per application, you might need to adjust its config in /etc/apparmor.d.
Check /var/log/syslog upon BIND start if it says something about “permission denied” about your selected logfile.
As Locutus may know, I believe this issue was in fact being caused by the security updates on our server not being able to overwrite temporary files, hence writing new temporary files that would not be deleted, clogging up the memory until the server crashed and knocked out the DNS server, along with a number of other services.
JamieCameron assisted me by logging in and updating our security software to fix the memory issue and the DNS server is no longer dropping out.
