Intermittent lookup failures and system limits?

Question

Hi,
on our deployment system we sometimes see intermitten failures of lookups. E.g. on the Webserver (Apache2.2) with a reverse proxy to an internal machine defined in /etc/hosts we will get "DNS Lookup failure", after a few reloads it works. Or restart of the apache helped, too. In our postgres mirroring script we have a similar problem. A host cannot be found even though the host is defined in /etc/hosts.
Since the problems are not 100% reproducable, I was thinking about some system limits (open files, processes?)
Did anybody run into a similar problem or has some hints where to look?
Christian

Answer 1

Hi,
I've been running a reverse proxied Webserver setup to an internal machine for over 4 months and not seen this problem. My reverse proxy instructions are located in /etc/apache2/sites/specifichostfile in the <IfModule modproxybalancer.c> module instructions.
I just use the internal IP to denote the machine not even bothering with a DNS entry. But an A record could be introduced to lookup the internal machine. If all the traffic on this domain goes to an internal machine you can specify the IP address in the WebService->General panel. In my case I am only diverting the /cgi-bin/ traffic in one case and Rails traffic in another, and so am more selective.
HTH,
Harry