run fsck prior to mount. should address hanging-mount failures on

[bootmanager.git] / source / steps / ValidateNodeInstall.py

diff --git a/source/steps/ValidateNodeInstall.py b/source/steps/ValidateNodeInstall.py
index 2e80714..278fea6 100644 (file)
--- a/source/steps/ValidateNodeInstall.py
+++ b/source/steps/ValidateNodeInstall.py
@@ -1,8 +1,16 @@
+#!/usr/bin/python2 -u
+
+# Copyright (c) 2003 Intel Corporation
+# All rights reserved.
+#
+# Copyright (c) 2004-2006 The Trustees of Princeton University
+# All rights reserved.
+
 import os
 
 from Exceptions import *
 import utils
-from systeminfo import systeminfo
+import systeminfo
 import compatibility
 import ModelOptions
 
@@ -47,6 +55,10 @@ def Run( vars, log ):
         
         NODE_MODEL_OPTIONS= vars["NODE_MODEL_OPTIONS"]
 
+        PARTITIONS= vars["PARTITIONS"]
+        if PARTITIONS == None:
+            raise ValueError, "PARTITIONS"
+
     except KeyError, var:
         raise BootManagerException, "Missing variable in vars: %s\n" % var
     except ValueError, var:
@@ -68,7 +80,7 @@ def Run( vars, log ):
         # simply creating an instance of this class and listing the system
         # block devices will make them show up so vgscan can find the planetlab
         # volume group
-        systeminfo().get_block_device_list()
+        systeminfo.get_block_device_list(vars, log)
 
         try:
             utils.sysexec( "vgscan", log )
@@ -81,12 +93,29 @@ def Run( vars, log ):
         utils.makedirs( SYSIMG_PATH )
 
         try:
-            utils.sysexec( "mount /dev/planetlab/root %s" % SYSIMG_PATH, log )
-            utils.sysexec( "mount /dev/planetlab/vservers %s/vservers" %
-                           SYSIMG_PATH, log )
+            # first run fsck to prevent fs corruption from hanging mount...
+            log.write( "fsck root file system\n" )
+            utils.sysexec("e2fsck -v -p %s" % (PARTITIONS["root"]),log)
+
+            log.write( "fsck vserver file system\n" )
+            utils.sysexec("e2fsck -v -p %s" % (PARTITIONS["vservers"]),log)
+        except BootManagerException, e:
+            log.write( "BootManagerException during fsck of /root and /vservers : %s\n" %
+                       str(e) )
+
+        try:
+            # then attempt to mount them
+            log.write( "mounting root file system\n" )
+            utils.sysexec("mount -t ext3 %s %s" % (PARTITIONS["root"],SYSIMG_PATH),log)
+
+            log.write( "mounting vserver partition in root file system\n" )
+            utils.sysexec("mount -t ext3 %s %s/vservers" % \
+                          (PARTITIONS["vservers"], SYSIMG_PATH), log)
+
+            log.write( "mounting /proc\n" )
             utils.sysexec( "mount -t proc none %s/proc" % SYSIMG_PATH, log )
         except BootManagerException, e:
-            log.write( "BootManagerException during vgscan/vgchange: %s\n" %
+            log.write( "BootManagerException during mount of /root, /vservers and /proc: %s\n" %
                        str(e) )
             return 0
 
@@ -94,22 +123,29 @@ def Run( vars, log ):
         vars['ROOT_MOUNTED']= 1
         
     
-    # get the kernel version
+    # check if the base kernel is installed
+    try:
+        os.stat("%s/boot/kernel-boot" % SYSIMG_PATH)
+        os.stat("%s/boot/initrd-boot" % SYSIMG_PATH)
+    except OSError, e:            
+        log.write( "FATAL: Couldn't locate base kernel.\n")                
+        return 0
+
+    # check if the model specified kernel is installed
     option = ''
     if NODE_MODEL_OPTIONS & ModelOptions.SMP:
         option = 'smp'
-
-    files = ("kernel-boot%s" % option, "initrd-boot%s" % option)
-    valid= 1
-    for filepath in files:
-        if not os.access("%s/boot/%s"%(SYSIMG_PATH,filepath),os.F_OK|os.R_OK):
-            log.write( "Node not properly installed:\n")
-            log.write( "\tmissing file /boot/%s\n" % filepath )
-            valid= 0
-    
-    if not valid:
-        return 0
-
+        try:
+            os.stat("%s/boot/kernel-boot%s" % (SYSIMG_PATH,option))
+            os.stat("%s/boot/initrd-boot%s" % (SYSIMG_PATH,option))
+        except OSError, e:
+            # smp kernel is not there; remove option from modeloptions
+            # such that the rest of the code base thinks we are just
+            # using the base kernel.
+            NODE_MODEL_OPTIONS = NODE_MODEL_OPTIONS & ~ModelOptions.SMP
+            vars["NODE_MODEL_OPTIONS"] = NODE_MODEL_OPTIONS
+            log.write( "WARNING: Couldn't locate smp kernel.\n")
+            
     # write out the node id to /etc/planetlab/node_id. if this fails, return
     # 0, indicating the node isn't a valid install.
     try:
@@ -118,9 +154,9 @@ def Run( vars, log ):
         node_id_file.write( str(NODE_ID) )
         node_id_file.close()
         node_id_file= None
-        log.write( "Updated /etc/planetlab/node_id" )
+        log.write( "Updated /etc/planetlab/node_id\n" )
     except IOError, e:
-        log.write( "Unable to write out /etc/planetlab/node_id" )
+        log.write( "Unable to write out /etc/planetlab/node_id\n" )
         return 0
 
     log.write( "Everything appears to be ok\n" )