Eskimo North


          [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

          Chat outage 6/30/99


          • To: outages-list@eskimo.com
          • Subject: Chat outage 6/30/99
          • From: Robert Dinse <nanook@eskimo.com>
          • Date: Fri, 2 Jul 1999 23:25:44 -0700 (PDT)
          • Resent-Date: Fri, 2 Jul 1999 23:25:49 -0700
          • Resent-From: outages-list@eskimo.com
          • Resent-Message-ID: <"zxysb1.0.U6.ipQVt"@mx1>
          • Resent-Sender: outages-list-request@eskimo.com

          
               We had an outage of chat.eskimo.com on 6/30/99.  I apologize for not
          posting this to now. 
          
               Chat crashed; it appears the crash was caused by some I/O errors on the
          swap partition leading to corruption of swap.
          
               When it tried to come back up, it bound to a ypserv running on our new SSL
          server that wasn't ready for use; had no database, and thus couldn't serve maps
          for eskimo.com domain.  Chat kept trying to bind to that machine and that
          prevented it from coming up fully. 
          
               It took me a while to figure out what was going on and kill the ypserv
          daemon on the new SSL server.
          
               The disks on chat are absolutely antique, so the fact that we get an
          occasional error isn't real suprising. 
          
               I have hardware on order that will replace both chat and isumataq (two sun
          LXs). These will be completely new boxes, new RAM, CPU, disk, everything.  The
          new hardware will be faster than the existing machines (50mhz version 8 Sparc
          vs. 16.7mhz [isumataq] and 25mhz [chat] version 7 Sparc). 
          
               The intent is, aside from saving a lot of electricity and reducing heat
          production, is to load Sparc Linux on the new machines and set them up with
          SunOS compatibility.  This will allow the existing binaries to run but it will
          also make it possible to compile and execute newer versions of eggdrop and
          newer MUD/Muck/Mush/Moo versions that won't compile in non-ANSI environments. 
          
               It's going to take some time to get these new machines setup and properly
          secured, particularly with a number of other projects occuring simultaneously.
          
          
          
          

          • Prev by Date: Sprint Newsfeed
          • Next by Date: Moving Modems
          • Prev by thread: Moving Modems
          • Next by thread: Sprint Newsfeed
          • Index(es):
            • Date
            • Thread