I have had users reporting intermittent times when my server is unreachable. It is hosting web/ftp/email. I haven't really experienced the problem myself until recently. After encountering the problem, I initially thought that maybe my server was being placed under a large load and unable to respond to requests. However, the highest values for CPU load in top was 0.01 for any of the time values. I then ran the bandwidth monitoring tool in Webmin for two days. The total over those two days was only 55 MB of traffic. So I think I have ruled out heavy CPU load and heavy network traffic load. I called my ISP last night just to ensure that they haven't throttled me or blocked any ports. Again, the answer was no it shouldn't be the problem. I have scanned my syslog pretty religiously looking for potential clues. I couldn't find any. I have a couple related to email (loop back to myself) but probably unrelated. Each time the server is unreachable through FTP or Web, I can still ping my gateway router. NSLOOKUP and DIG resolve to the correct IP address for the gateway. So it appears to be something on my end. I was ready to give up not knowing where to look next. Today however, I noticed that it isn't unreachable by all clients at the same time. For example, I wasn't able to pull up any websites on my work computer. They just time out. But I was able to browse to the site on my web enabled phone. My buddy did the same and was able to connect. After a few minutes, his phone started timing out on web requests, but mine never did. It was still pulling pages from the server. I am at a loss here. I don't know what could be causing this problem. Thanks.
Are there any errors in your logs? Did you check if the RAM is ok (e.g. with memtest)? To solve this (unrelated) problem, just add the domains/hostname to mydestination in /etc/postfix/main.cf.
I haven't had to do a memory test before. My quick search online appears that I have to perform the test from the Ubuntu LiveCD. Is that true? If so, how can I determine what the memory usage and any errors are during normal production? On the unrelated topic of the mail loop...Should I set the mydestination if I am using virtual hosting through a MySQL backend? I thought I read somewhere that I should leave the setting blank in this scenario. Thanks.
Yes. Not sure how to find out about memory errors on a running system, but regarding the memory usage, take a look at Code: top or Code: cat /proc/meminfo If you're getting these errors for domains that you don't use for virtual accounts, then you can add them to mydestination.
So I took the server down for two hours last night and ran memtest on the machine. It did four passes in two hours with zero errors. I am thinking I might just do a complete rebuild. I have a new drive I want to put in any way and this machine has been a test machine for a lot of different software installations. I think I will plan on using the Perfect Server How-To for Ubuntu 7.10.
