initscripts/pl_boot

   1 #!/bin/bash
   2
   3 . /tmp/planet.cnf
   4
   5 # Run gpg once to create default options
   6 GNUPGHOME=/root
   7 export GNUPGHOME
   8 /usr/bin/gpg --yes 2>/dev/null </dev/null
   9
  10 # if this file is present, cancel the boot (exit this script)
  11 CANCEL_BOOT_FLAG=/tmp/CANCEL_BOOT
  12
  13 # how many times to fail in attempting to contact primary server
  14 # before falling back to original. if the backup fails this many times
  15 # too, then the process is repeated started with the primary server
  16 ATTEMPTS_BEFORE_BACKUP=3
  17
  18 # where all the configuration files for contacting
  19 # the boot server are stored
  20 BOOT_DIR=/usr/boot/
  21
  22 # get the server we are going to be contacting
  23 BOOT_SERVER=`cat $BOOT_DIR/boot_server`
  24 BOOT_SERVER_PORT=`cat $BOOT_DIR/boot_server_port`
  25
  26 # the file to request from the boot server
  27 BOOT_SERVER_PATH=`cat $BOOT_DIR/boot_server_path`
  28
  29 # location of the cacert for this boot server
  30 BOOT_SERVER_CACERT=$BOOT_DIR/cacert.pem
  31
  32 # location of the gpg key ring to verify scripts
  33 BOOT_SERVER_GPG_KEYRING=$BOOT_DIR/pubring.gpg
  34
  35 # get the backup server we are going to be contacting
  36 BACKUP_BOOT_SERVER=`cat $BOOT_DIR/backup/boot_server`
  37 BACKUP_BOOT_SERVER_PORT=`cat $BOOT_DIR/backup/boot_server_port`
  38
  39 # the file to request from the backup boot server
  40 BACKUP_BOOT_SERVER_PATH=`cat $BOOT_DIR/backup/boot_server_path`
  41
  42 # location of the cacert for the backup boot server
  43 BACKUP_BOOT_SERVER_CACERT=$BOOT_DIR/backup/cacert.pem
  44
  45 # location of the gpg key ring for backup server to verify scripts
  46 BACKUP_BOOT_SERVER_GPG_KEYRING=$BOOT_DIR/backup/pubring.gpg
  47
  48 # location of a file containing this boot cd version
  49 BOOT_VERSION_FILE=/pl_version
  50
  51 # the locations of the downloaded scripts
  52 UNVERIFIED_SCRIPT=/tmp/bootscript.gpg
  53 VERIFIED_SCRIPT=/tmp/bootscript
  54
  55
  56 # --------------------------
  57
  58
  59 # now, contact the boot server, run the script, and do it over again.
  60 contact_count=0
  61
  62 # set to one when we are trying to contact backup server
  63 on_backup_server=0
  64
  65 # start out contacting the primary servers
  66 CONNECT_BOOT_SERVER=$BOOT_SERVER
  67 CONNECT_BOOT_SERVER_PORT=$BOOT_SERVER_PORT
  68 CONNECT_BOOT_SERVER_PATH=$BOOT_SERVER_PATH
  69 CONNECT_BOOT_SERVER_GPG_KEYRING=$BOOT_SERVER_GPG_KEYRING
  70 CONNECT_BOOT_SERVER_CACERT=$BOOT_SERVER_CACERT
  71
  72 while : ; do
  73
  74     if [[ -f $CANCEL_BOOT_FLAG ]]; then
  75         echo $(date "+%H:%M:%S") " pl_boot: got request to cancel boot, exiting"
  76         exit 0
  77     fi
  78
  79     if [[ $contact_count -ge $ATTEMPTS_BEFORE_BACKUP ]]; then
  80
  81         contact_count=0
  82
  83         if [[ $on_backup_server == 1 ]]; then
  84             echo $(date "+%H:%M:%S") " pl_boot: failed to contact backup server, trying primary $BOOT_SERVER"
  85
  86             on_backup_server=0
  87
  88             CONNECT_BOOT_SERVER=$BOOT_SERVER
  89             CONNECT_BOOT_SERVER_PORT=$BOOT_SERVER_PORT
  90             CONNECT_BOOT_SERVER_PATH=$BOOT_SERVER_PATH
  91             CONNECT_BOOT_SERVER_GPG_KEYRING=$BOOT_SERVER_GPG_KEYRING
  92             CONNECT_BOOT_SERVER_CACERT=$BOOT_SERVER_CACERT
  93         else
  94             echo $(date "+%H:%M:%S") " pl_boot: failed to contact primary server, trying backup $BACKUP_BOOT_SERVER"
  95
  96             on_backup_server=1
  97
  98             CONNECT_BOOT_SERVER=$BACKUP_BOOT_SERVER
  99             CONNECT_BOOT_SERVER_PORT=$BACKUP_BOOT_SERVER_PORT
 100             CONNECT_BOOT_SERVER_PATH=$BACKUP_BOOT_SERVER_PATH
 101             CONNECT_BOOT_SERVER_GPG_KEYRING=$BACKUP_BOOT_SERVER_GPG_KEYRING
 102             CONNECT_BOOT_SERVER_CACERT=$BACKUP_BOOT_SERVER_CACERT
 103         fi
 104     fi
 105
 106     if [[ $contact_count != 0 ]]; then
 107
 108         if [[ $on_backup_server == 1 ]]; then
 109             echo $(date "+%H:%M:%S") " pl_boot: attempting to fetch script from backup server in 30s"
 110         else
 111             echo $(date "+%H:%M:%S") " pl_boot: attempting to fetch script from primary server in 30s"
 112         fi
 113         /bin/sleep 30
 114     fi
 115
 116     # assemble the curl transaction
 117     CURL_CMD="/usr/bin/curl \
 118         --connect-timeout 60 \
 119         --max-time 600 \
 120         --form version=<$BOOT_VERSION_FILE \
 121         --form cmdline=</proc/cmdline \
 122         --form uptime=</proc/uptime \
 123         --form ifconfig=</tmp/ifconfig \
 124         --form nonce=</tmp/nonce \
 125         --location \
 126         --output $UNVERIFIED_SCRIPT \
 127         --sslv3  \
 128         --silent \
 129         --show-error \
 130         --fail \
 131         --stderr /tmp/curl_errors \
 132         --cacert $CONNECT_BOOT_SERVER_CACERT \
 133    https://$CONNECT_BOOT_SERVER:$CONNECT_BOOT_SERVER_PORT/$CONNECT_BOOT_SERVER_PATH"
 134
 135     # assemble the gpg command line
 136     GPG_CMD="/usr/bin/gpg \
 137         --no-default-keyring \
 138         --keyring $CONNECT_BOOT_SERVER_GPG_KEYRING \
 139         --output $VERIFIED_SCRIPT \
 140         --always-trust \
 141         --decrypt $UNVERIFIED_SCRIPT"
 142
 143     echo $(date "+%H:%M:%S") " pl_boot: generating new nonce"
 144     /usr/bin/head --bytes=32 /dev/urandom | \
 145     /usr/bin/od -tx1 -An --width=32 | \
 146     /bin/sed 's/ //g' > /tmp/nonce
 147
 148     echo $(date "+%H:%M:%S") " pl_boot: fetching script from boot server $CONNECT_BOOT_SERVER"
 149     ((contact_count++))
 150     rm -f $UNVERIFIED_SCRIPT
 151     $CURL_CMD
 152     curl_err=$?
 153     if [ $curl_err -ne 0 ]; then
 154         echo $(date "+%H:%M:%S") " pl_boot: curl request failed with error $curl_err:"
 155         cat /tmp/curl_errors
 156         echo
 157         if [ -n "$DISCONNECTED_OPERATION" ]; then
 158             mkdir /tmp/boot-media
 159             mount -U "$DISCONNECTED_OPERATION" /tmp/boot-media
 160             cp /tmp/boot-media/bootscript.gpg $UNVERIFIED_SCRIPT
 161             umount /tmp/boot-media
 162             rmdir /tmp/boot-media
 163         else
 164                         case $curl_err in
 165                                 6)
 166                                 echo $(date "+%H:%M:%S") " This error likely indicates a networking configuration error. "
 167                                 echo $(date "+%H:%M:%S") " Please, check whether you can ping this machine.  If you can, "
 168                                 echo $(date "+%H:%M:%S") " we recommend checking your DNS settings.  If you cannot, then "
 169                                         echo $(date "+%H:%M:%S") " please double check your network settings registered at PLC and "
 170                                         echo $(date "+%H:%M:%S") " stored on this Boot Image."
 171                                         for file in `ls /etc/sysconfig/network-scripts/ifcfg-eth*` ; do
 172                                                 echo $(date "+%H:%M:%S") $file :
 173                                                 cat $file
 174                                         done
 175                                 ;;
 176                                 60)
 177                                 echo $(date "+%H:%M:%S") " This error likely indicates that the hardware clock is not set "
 178                                         echo $(date "+%H:%M:%S") " to GMT.  The result is that authentication between the local and "
 179                                         echo $(date "+%H:%M:%S") " remote site fails.  Please double check this machine's system "
 180                                         echo $(date "+%H:%M:%S") " clock, and set it to GMT in the BIOS. If after rebooting the same "
 181                                         echo $(date "+%H:%M:%S") " error occurs, please report the situation to support@planet-lab.org "
 182                                 echo $(date "+%H:%M:%S") " with as much detail as possible."
 183                                 ;;
 184                                 *)
 185                                 echo $(date "+%H:%M:%S") " The best-practice approach for handling this error is not yet "
 186                                 echo $(date "+%H:%M:%S") " documented.  Please report this error to support@planet-lab.org "
 187                                 echo $(date "+%H:%M:%S") " with as much detail as possible."
 188                                 ;;
 189                         esac
 190             continue
 191         fi
 192     elif [ -n "$DISCONNECTED_OPERATION" ]; then
 193         mkdir /tmp/boot-media
 194         mount -U "$DISCONNECTED_OPERATION" /tmp/boot-media
 195         cp $UNVERIFIED_SCRIPT /tmp/boot-media
 196         umount /tmp/boot-media
 197         rmdir /tmp/boot-media
 198     fi
 199
 200     echo $(date "+%H:%M:%S") " pl_boot: verifying downloaded script"
 201     rm -f $VERIFIED_SCRIPT
 202     $GPG_CMD 2> /tmp/gpg_errors
 203     if [ $? -ne 0 ]; then
 204         echo $(date "+%H:%M:%S") " pl_boot: failed to verify file:"
 205         cat /tmp/gpg_errors
 206         echo
 207         continue
 208     fi
 209     echo $(date "+%H:%M:%S") " pl_boot: decrypted and verified script succesfully"
 210
 211     echo $(date "+%H:%M:%S") " pl_boot: handing control to download script"
 212     rm -f $UNVERIFIED_SCRIPT
 213     chmod +x $VERIFIED_SCRIPT
 214     $VERIFIED_SCRIPT
 215
 216     echo $(date "+%H:%M:%S") " pl_boot: downloaded script has returned"
 217 done
 218
 219 echo $(date "+%H:%M:%S") " pl_boot: automatic boot process canceled by user"