Added set_dlimit and get_dlimit to be used by NM to enforce disk limits.
[util-vserver.git] / python / vserver.py
index feeb026..2d72bd0 100644 (file)
@@ -4,11 +4,17 @@ import errno
 import fcntl
 import os
 import re
+import sys
+import time
+import traceback
 
+import mountimpl
 import linuxcaps
 import passfdimpl
+import utmp
+import vserverimpl, vduimpl
 
-from vserver_vars import *
+from util_vserver_vars import *
 
 CAP_SAFE = (linuxcaps.CAP_CHOWN |
             linuxcaps.CAP_DAC_OVERRIDE |
@@ -25,21 +31,24 @@ CAP_SAFE = (linuxcaps.CAP_CHOWN |
             linuxcaps.CAP_SYS_PTRACE)
 
 #
-# XXX - these are the flags taken from chcontext.c, but they don't match
-# up with those apparently used by the kernel
+# these are the flags taken from the kernel linux/vserver/legacy.h
 #
 FLAGS_LOCK = 1
-FLAGS_SCHED = 2
+FLAGS_SCHED = 2  # XXX - defined in util-vserver/src/chcontext.c
 FLAGS_NPROC = 4
 FLAGS_PRIVATE = 8
-FLAGS_FAKEINIT = 16
+FLAGS_INIT = 16
 FLAGS_HIDEINFO = 32
 FLAGS_ULIMIT = 64
+FLAGS_NAMESPACE = 128
 
 
               
 class VServer:
 
+    INITSCRIPTS = [('/etc/rc.vinit', 'start'),
+                   ('/etc/rc.d/rc', '%(runlevel)d')]
+
     def __init__(self, name):
 
         self.name = name
@@ -53,7 +62,7 @@ class VServer:
         if "nproc" in flags:
             self.flags |= FLAGS_NPROC
         self.remove_caps = ~CAP_SAFE
-        print "%x %x" % (self.flags, ~self.remove_caps)
+        self.ctx = int(self.config["S_CONTEXT"])
 
     config_var_re = re.compile(r"^ *([A-Z_]+)=(.*)\n?$", re.MULTILINE)
 
@@ -68,14 +77,36 @@ class VServer:
             config[key] = val.strip('"')
         return config
 
-    def open(self, filename, mode = "r"):
+    def __do_chroot(self):
+
+        return os.chroot("%s/%s" % (VROOTDIR, self.name))
+
+    def set_dlimit(self, blocktotal):
+        path = "%s/%s" % (VROOTDIR, self.name)
+        inodes, blockcount, size = vduimpl.vdu(path)
+        vserverimpl.setdlimit(path, self.ctx, blockcount>>1, blocktotal, inodes, -1, 2)
+
+    def get_dlimit(self):
+        path = "%s/%s" % (VROOTDIR, self.name)
+        try:
+            blocksused, blocktotal, inodesused, inodestotal, reserved = \
+                        vserverimpl.getdlimit(path,self.ctx)
+        except OSError, ex:
+            if ex.errno == 3:
+                # get here if no vserver disk limit has been set for xid
+                # set blockused to -1 to indicate no limit
+                blocktotal = -1
+
+        return blocktotal
+
+    def open(self, filename, mode = "r", bufsize = -1):
 
         (sendsock, recvsock) = passfdimpl.socketpair()
         child_pid = os.fork()
         if child_pid == 0:
             try:
                 # child process
-                os.chroot("%s/%s" % (VROOTDIR, self.name))
+                self.__do_chroot()
                 f = open(filename, mode)
                 passfdimpl.sendmsg(f.fileno(), sendsock)
                 os._exit(0)
@@ -123,4 +154,128 @@ class VServer:
         if not fd:
             throw()
 
-        return os.fdopen(fd)
+        return os.fdopen(fd, mode, bufsize)
+
+    def __do_chcontext(self, state_file = None):
+
+        vserverimpl.chcontext(self.ctx, self.remove_caps)
+        if not state_file:
+            return
+        print >>state_file, "S_CONTEXT=%d" % self.ctx
+        print >>state_file, "S_PROFILE=%s" % self.config.get("S_PROFILE", "")
+        state_file.close()
+
+    def __prep(self, runlevel, log):
+
+        """ Perform all the crap that the vserver script does before
+        actually executing the startup scripts. """
+
+        # remove /var/run and /var/lock/subsys files
+        # but don't remove utmp from the top-level /var/run
+        RUNDIR = "/var/run"
+        LOCKDIR = "/var/lock/subsys"
+        filter_fn = lambda fs: filter(lambda f: f != 'utmp', fs)
+        garbage = reduce((lambda (out, ff), (dir, subdirs, files):
+                          (out + map((dir + "/").__add__, ff(files)),
+                           lambda fs: fs)),
+                         list(os.walk(RUNDIR)),
+                         ([], filter_fn))[0]
+        garbage += filter(os.path.isfile, map((LOCKDIR + "/").__add__,
+                                              os.listdir(LOCKDIR)))
+        for f in garbage:
+            os.unlink(f)
+
+        # set the initial runlevel
+        f = open(RUNDIR + "/utmp", "w")
+        utmp.set_runlevel(f, runlevel)
+        f.close()
+
+        # mount /proc and /dev/pts
+        self.__do_mount("none", "/proc", "proc")
+        # XXX - magic mount options
+        self.__do_mount("none", "/dev/pts", "devpts", 0, "gid=5,mode=0620")
+
+    def __do_mount(self, *mount_args):
+
+        try:
+            mountimpl.mount(*mount_args)
+        except OSError, ex:
+            if ex.errno == errno.EBUSY:
+                # assume already mounted
+                return
+            raise ex
+
+    def enter(self):
+
+        state_file = open("/var/run/vservers/%s.ctx" % self.name, "w")
+        self.__do_chroot()
+        self.__do_chcontext(state_file)
+
+    def start(self, wait, runlevel = 3):
+
+        child_pid = os.fork()
+        if child_pid == 0:
+            # child process
+            try:
+                # get a new session
+                os.setsid()
+
+                # open state file to record vserver info
+                state_file = open("/var/run/vservers/%s.ctx" % self.name, "w")
+
+                # use /dev/null for stdin, /var/log/boot.log for stdout/err
+                os.close(0)
+                os.close(1)
+                os.open("/dev/null", os.O_RDONLY)
+                self.__do_chroot()
+                log = open("/var/log/boot.log", "w", 0)
+                os.dup2(1, 2)
+
+                print >>log, ("%s: starting the virtual server %s" %
+                              (time.asctime(time.gmtime()), self.name))
+
+                # perform pre-init cleanup
+                self.__prep(runlevel, log)
+
+                # execute each init script in turn
+                # XXX - we don't support all scripts that vserver script does
+                cmd_pid = 0
+                for cmd in self.INITSCRIPTS + [None]:
+                    # wait for previous command to terminate, unless it
+                    # is the last one and the caller has specified to wait
+                    if cmd_pid and (cmd != None or wait):
+                        try:
+                            os.waitpid(cmd_pid, 0)
+                        except:
+                            print >>log, "error waiting for %s:" % cmd_pid
+                            traceback.print_exc()
+
+                    # end of list
+                    if cmd == None:
+                        os._exit(0)
+
+                    # fork and exec next command
+                    cmd_pid = os.fork()
+                    if cmd_pid == 0:
+                        try:
+                            # enter vserver context
+                            self.__do_chcontext(state_file)
+                            arg_subst = { 'runlevel': runlevel }
+                            cmd_args = [cmd[0]] + map(lambda x: x % arg_subst,
+                                                      cmd[1:])
+                            print >>log, "executing '%s'" % " ".join(cmd_args)
+                            os.execl(cmd[0], *cmd_args)
+                        except:
+                            traceback.print_exc()
+                            os._exit(1)
+                    else:
+                        # don't want to write state_file multiple times
+                        state_file = None
+
+            # we get here due to an exception in the top-level child process
+            except Exception, ex:
+                traceback.print_exc()
+            os._exit(0)
+
+        # parent process
+        return child_pid