HCoop Git - hcoop/zz_old/ikiwiki/blob - OnSiteStuff.mdwn

   1 This page is a checklist for ntk, so that he remembers what to bring with him on the next on-site visit (or future on-site visits, for purchases).
   2
   3  * Philips screwdriver
   4  * 2nd new hard drive
   5  * Burned RIP CD (from [http://www.tux.org/pub/people/kent-robotti/looplinux/rip/RIPLinux-2.9.iso])
   6  * Burned Debian Etch install CD
   7  * Netra needs to be mounted using the appropriate screws.  The Netra apparently does not require rails.
   8  * 9V+ type round adapter power cable for the new KVM switch
   9
  10 Here is the plan for salvaging mire.
  11
  12  * Boot off the RIP CD.
  13   * If without RIP CD but with Knoppix, boot using this cheat-code: {{{knoppix-txt 3}}}  This should use {{{vga=normal}}} as a kernel parameter and boot into runlevel 3, command line only.  The point is to allow VNC to function properly as using the framebuffer may cause display issues with VNC.
  14  * Mount /dev/sda1 on /mnt (remember, this is the old /dev/sdb, when it is the only drive in the machine).
  15  * mwolson will:
  16    1. selectively copy things off of /etc onto deleuze (he will use git for this to clone his existing /etc repository, which can ensure that the contents are not corrupt)
  17    2. save the output of df, fdisk -l /dev/sda and dpkg -l
  18    3. copy the entire /etc directory to deleuze (for reference, in case mwolson missed something important in the earlier copying run)
  19    4. fsck /dev/sda*
  20  * Turn machine off, take out the hard drive, put in one new hard drive only, put the Debian Etch install CD in the CD drive.
  21  * Install Debian on the new drive, and boot from it.
  22    * Run bonnie++ (preferable) or dd to stress test the new HD and io subsystem, before wasting time installing too many other packages and configuring, etc.  Watch for errors during test.
  23      * bonnie -d /test/ -s 0.1 -n 1 -r 252 -u root
  24      * bonnie -d /test/ -s 0.1 -n 10 -r 252 -u root
  25    * Start copying over saved settings and packages to it.
  26    * ntk is free to leave at this point if he wants.
  27    * If the machine stays up, and does not lock up for a day or two:
  28      * We will assume that both of the old hard drives were bad, instead of just the old /dev/sda.
  29      * We will install the second hard drive.
  30      * We will bind the two hard drives together via RAID.
  31    * Otherwise, if the machine does lock up:
  32      * Obviously something else is wrong.
  33      * Look into various hardware we can try to replace, if there is some sort of SCSI card.
  34      * If the motherboard is bad, we will probably have to scrap mire, unless someone has other plans.