Results 1 to 5 of 5
  1. #1

    Linux: Remote Administration

    I'm thinking about putting a server in a colo where the Datacenter is a 4 hour drive.

    My question is: "Do Linux machines running LAMP hang / freeze often so that you cannot force a reboot remotely ?" - "Does it even happen on Linux boxes?"

    I have a lot of experience with Windows machines but none with Linux. With a Windows machine I for sure would have to purchase a DRAC/LILO board.

    Unfortunately, I don't have any production experience with Linux so I don't know the risk of having to go on site because Linux decided to completely freeze making remote management impossible. Of course this is not included any hardware outages.

    Many thanks !


  2. #2
    Join Date
    Feb 2003
    North Hollywood, CA
    Quote Originally Posted by endymon
    My question is: "Do Linux machines running LAMP hang / freeze often so that you cannot force a reboot remotely ?" - "Does it even happen on Linux boxes?"

    As so remote administration theres telnet/ssh CLi, or VNC GUI or CLi and a bunch of others.

    If the box kernel paniced a reboot is in order. Finding the true caused might be a task (bad mem, HD, bad compile, box is warm etc etc.)

    Some colo's offer a remote reboot port that shuts the power off at the port then in a few seconds gives the power back. If you configure the bios correct it should come back online.
    Remote Hands and Your Local Tech for the Los Angeles area.

    (310) 573-8050 - LinkedIn

  3. #3
    Join Date
    Jan 2003
    Lake Arrowhead, CA
    The short answer is simply no. If system lockups are a week to week or even month to month issue, then the system in question is either built on an inherently unstable hardware platform or is not running 'production quality' software.

    The long answer is it depends on the greater combination of hardware and software involved. Stable releases of Apache, Mysql and PHP alone aren't going to "decide to freeze" running on stable hardware under ANY server OS. We have two Windows 2000 servers (with Apache, Mysql, PHP and more) which have never needed an unscheduled reboot for five years running. They aren't shared servers, but the point remains that high quality hardware and software should NEVER hang up until the hardware simply gets old and starts deteriorating. The same holds true with Linux. It's only when we move to a new server platform or OS version that we tend to experience lockups. Then we find out what the issues are, resolve them (sometimes going back to older, proven hardware, a different kernel, etc.) and that tends to be it.

    In an industry where hardware and software designs change at a very high rate of speed, you are forced with a very basic choice: Are you an early adopter or do you chose long term stability? If you go with proven choices and don't cut corners, you can avoid nearly all hang/freeze situations on non-test platforms.

    That's not to say you don't want remote power or on-site personnel available for those unknowns. Of course you do, because hardware does simply fail from time to time. But you certainly shouldn't be using it frequently for a single server.
    Last edited by SROHost; 01-07-2007 at 11:55 AM.
    Stability, redundancy and peace of mind

  4. #4
    Join Date
    Oct 2004
    Southwest UK
    A lockup only has to happen once to ruin your day. So while linux is very stable (assuming you're not running some bleeding edge software), it can happen, but in such cases a reboot will sort it out and you'll probably never even figure out what the cause was. (cosmic rays perhaps. No, really. I'm serious)

    Hardware failures are something else though, as a reboot will only fix a overheated server for a few minutes before it shuts down, and if a drive has a broken MBR sector, rebooting a working server will not bring it back up.

    Generally, a host that offers free reboots is a good thing if you can't get a remote power-socket reboot port. FYI, I've been colo-ed for 2 years with 2 boxes and I've been onsite 2 times (plus the initial drive down there). Both times due to hardware failures (a HDD and a CPU fan)
    Do not meddle in the affairs of Dragons, for you are crunchy and taste good.

  5. #5
    Join Date
    Nov 2006
    College Station, TX
    Most colo providers offer a 'remote hands' service where you can call 24/7 and someone on their staff will run to the server and power cycle it for you. Alternately, purchase an HP server with Integrated Lights Out or some other such remote power cycling hardware... but I've been running linux boxes on the LAMP stack for years and only had one freeze in an unrecoverable way for no reason at all once.

    Now, development boxes where you're playing with new software is another matter. If you'll have untried or untested software running on it (including PHP scripts) or it's a shared hosting box, then definitely have a way to power cycle it remotely, whether that's a telnet-enabled power strip or a remote board.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts