1

My server crashed 3 times, (2 times less than 24 hours ago), and once 1 week ago. Each time, the server was not pingable, until the host rebooted the machine. It's a dedicated server at ServerPronto with 4 GB of RAM, dual or quad cpu with 3.1 ghz, located in Florida, i think. More then enough power to handle the load with tons of cpu to spare. Innodb buffer pool has enough memory (1.3gb) to have the whole mysql table in memory with extra space. APC has more then enough memory too.

I have New Relic installed on my server. Each of the 3 times, I checked CPU usage, Memory usage, Disk space, networking IO, and all of them were completely stable. I looked at the response times for requests and they were all minimally normal. My server got 500 requests per minute, durring the hour that led up to the crash: this is very normal for my server. Then all of the sudden, without any spikes at all, the requests per minute went down to 0! The load average was 0.05 (very very small, but normal for my server), memory usage was at 30%, also normal, network IO was at 400 KB / sec, also normal. Disk IO was at 1-2%, also normal. The log file must have been working because I see the standard warnings in the apache log files, but absolutely no errors before the crash.

The last shutdown, occurred during a time when my traffic gets a minimal amount of web traffic.

Furthermore, there are no cron jobs running at the time of the incidents. There are no errors or warning in the error logs, PHP/mysql logs at the time of the shutdown. No IDEs or SSH connections were open to the server at the time of the shutdown.

Each time it goes down, as soon as the restart happens, every works again.

I've been running the same software, for 2.5 years on two of my previous hosts and didn't have this type of problem. Also, I haven't made any code changes in the last month (for the app currently running).

When I first set up this machine with serverPronto about 1 month ago, they had some hardware issue with cable to my machine: Not a good start. But they fixed that, and I didn't have any problems for about 4 weeks (except for 1 ten minute outage which did not require a server restart)

Considering all this information, Does this mean that ServerPronto (my host) is having a hardware issue/issue on their side currently? or is there another possibility.

Pascal Aschwanden
  • 41
  • 2
  • 1
    Look at the machine's console at the time of the crash to see any panic messages. If the provider doesn't allow you access to the console (e.g. via an IP-based KVM switch) then it's time to go somewhere else. – Michael Hampton Dec 31 '13 at 18:58
  • they checked the console for me and said they could not find any error messages. – Pascal Aschwanden Dec 31 '13 at 19:38
  • Typical - for something low level (driver, hardware issue). Not everything leaves the computer in a state to still write to the disc in a controlled fashion. – TomTom Dec 31 '13 at 20:18

1 Answers1

4

When I first set up this machine with serverPronto about 1 month ago,

Bad news. They STILL have hardware issues. Simple like that.

If you have the correct drivers installed then the answer is trivial - there is broken hardware involved.

TomTom
  • 51,649
  • 7
  • 54
  • 136