move clean_policy.py into monitor package

[monitor.git] / bootman.py

diff --git a/bootman.py b/bootman.py
index a278afe..981a911 100755 (executable)
--- a/bootman.py
+++ b/bootman.py
@@ -20,11 +20,11 @@ from sets import Set
 import ssh.pxssh as pxssh
 import ssh.fdpexpect as fdpexpect
 import ssh.pexpect as pexpect
-from unified_model import *
+from monitor.model import *
 from emailTxt import mailtxt
 from nodeconfig import network_config_to_str
 import traceback
-import monitorconfig
+import config
 
 import signal
 class Sopen(subprocess.Popen):
@@ -34,11 +34,7 @@ class Sopen(subprocess.Popen):
 #from Rpyc import SocketConnection, Async
 from Rpyc import SocketConnection, Async
 from Rpyc.Utils import *
-
-def get_fbnode(node):
-       fb = database.dbLoad("findbad")
-       fbnode = fb['nodes'][node]['values']
-       return fbnode
+fb = None
 
 class NodeConnection:
        def __init__(self, connection, node, config):
@@ -204,7 +200,7 @@ class PlanetLabSession:
                args['port'] = self.port
                args['user'] = 'root'
                args['hostname'] = self.node
-               args['monitordir'] = monitorconfig.MONITOR_SCRIPT_ROOT
+               args['monitordir'] = config.MONITOR_SCRIPT_ROOT
                ssh_port = 22
 
                if self.nosetup:
@@ -311,7 +307,7 @@ def reboot(hostname, config=None, forced_action=None):
 
        # NOTE: Nothing works if the bootcd is REALLY old.
        #       So, this is the first step.
-       fbnode = get_fbnode(hostname)
+       fbnode = FindbadNodeRecord.get_latest_by(hostname=hostname).to_dict()
        if fbnode['category'] == "OLDBOOTCD":
                print "...NOTIFY OWNER TO UPDATE BOOTCD!!!"
                args = {}
@@ -359,7 +355,6 @@ def reboot(hostname, config=None, forced_action=None):
                except:
                        print traceback.print_exc()
                        return False
-                       
 
        if forced_action == "reboot":
                conn.restart_node('rins')
@@ -503,6 +498,7 @@ def reboot(hostname, config=None, forced_action=None):
                        ('nodehostname' , 'Configured node hostname does not resolve'),
                        ('implementerror', 'Implementation Error'),
                        ('readonlyfs'   , '[Errno 30] Read-only file system'),
+                       ('baddisk'      , "IOError: [Errno 13] Permission denied: '/tmp/mnt/sysimg//vservers/\w+/etc/hosts'"),
                        ('noinstall'    , 'notinstalled'),
                        ('bziperror'    , 'bzip2: Data integrity error when decompressing.'),
                        ('noblockdev'   , "No block devices detected."),
@@ -512,6 +508,7 @@ def reboot(hostname, config=None, forced_action=None):
                        ('hardwarerequirefail' , 'Hardware requirements not met'),
                        ('mkfsfail'         , 'while running: Running mkfs.ext2 -q  -m 0 -j /dev/planetlab/vservers failed'),
                        ('nofilereference', "No such file or directory: '/tmp/mnt/sysimg//vservers/.vref/planetlab-f8-i386/etc/hosts'"),
+                       ('kernelcopyfail', "cp: cannot stat `/tmp/mnt/sysimg/boot/kernel-boot': No such file or directory"),
                        ('chrootfail'   , 'Running chroot /tmp/mnt/sysimg'),
                        ('modulefail'   , 'Unable to get list of system modules'),
                        ('writeerror'   , 'write error: No space left on device'),
@@ -539,11 +536,11 @@ def reboot(hostname, config=None, forced_action=None):
        #  By using the sequence identifier, we guarantee that there will be no
        #  frequent loops.  I'm guessing there is a better way to track loops,
        #  though.
-       if not config.force and pflags.getRecentFlag(s):
-               pflags.setRecentFlag(s)
-               pflags.save() 
-               print "... flag is set or it has already run recently. Skipping %s" % node
-               return True
+       #if not config.force and pflags.getRecentFlag(s):
+       #       pflags.setRecentFlag(s)
+       #       pflags.save() 
+       #       print "... flag is set or it has already run recently. Skipping %s" % node
+       #       return True
 
        sequences = {}
 
@@ -581,7 +578,13 @@ def reboot(hostname, config=None, forced_action=None):
                        "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done",
                        "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-exception-mkfsfail-update-debug-done",
                        "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done",
+                       "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-exception-chrootfail-update-debug-done",
+                       "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done",
+                       "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done",
                        "bminit-cfg-auth-getplc-installinit-validate-exception-noinstall-update-debug-done",
+                       # actual solution appears to involve removing the bad files, and
+                       # continually trying to boot the node.
+                       "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-update-debug-done",
                        ]:
                sequences.update({n : "restart_bootmanager_rins"})