def cleanup(nodes, cleantmp=False):
hadError = False
util.output("cleaning up nodes ...")
result = isRunning(nodes)
running = [node for (node, on) in result if on]
notrunning = [node for (node, on) in result if not on]
results1 = execute.rmdirs([(n, n.cwd()) for n in notrunning])
results2 = execute.mkdirs([(n, n.cwd()) for n in notrunning])
if nodeFailed(results1) or nodeFailed(results2):
hadError = True
for node in notrunning:
node.clearCrashed()
for node in running:
util.output(" %s is still running, not cleaning work directory" % node.name)
if cleantmp:
results3 = execute.rmdirs([(n, config.Config.tmpdir) for n in running + notrunning])
results4 = execute.mkdirs([(n, config.Config.tmpdir) for n in running + notrunning])
if nodeFailed(results3) or nodeFailed(results4):
hadError = True
return not hadError
def cleanup(nodes, cleantmp=False):
util.output("cleaning up nodes ...")
result = isRunning(nodes)
running = [node for (node, on) in result if on]
notrunning = [node for (node, on) in result if not on]
execute.rmdirs([(n, n.cwd()) for n in notrunning])
execute.mkdirs([(n, n.cwd()) for n in notrunning])
for node in notrunning:
node.clearCrashed();
for node in running:
util.output(" %s is still running, not cleaning work directory" % node.name)
if cleantmp:
execute.rmdirs([(n, config.Config.tmpdir) for n in running + notrunning])
execute.mkdirs([(n, config.Config.tmpdir) for n in running + notrunning])
def cleanup(nodes, cleantmp=False):
util.output("cleaning up nodes ...")
result = isRunning(nodes)
running = [node for (node, on) in result if on]
notrunning = [node for (node, on) in result if not on]
execute.rmdirs([(n, n.cwd()) for n in notrunning])
execute.mkdirs([(n, n.cwd()) for n in notrunning])
for node in notrunning:
node.clearCrashed()
for node in running:
util.output(" %s is still running, not cleaning work directory" %
node.name)
if cleantmp:
execute.rmdirs([(n, config.Config.tmpdir)
for n in running + notrunning])
execute.mkdirs([(n, config.Config.tmpdir)
for n in running + notrunning])
def _startNodes(nodes):
results = []
filtered = []
# Ignore nodes which are still running.
for (node, isrunning) in isRunning(nodes):
if not isrunning:
filtered += [node]
util.output("starting %s ..." % node.name)
else:
util.output("%s still running" % node.name)
nodes = filtered
# Generate crash report for any crashed nodes.
crashed = [node for node in nodes if node.hasCrashed()]
_makeCrashReports(crashed)
# Make working directories.
dirs = [(node, node.cwd()) for node in nodes]
nodes = []
for (node, success) in execute.mkdirs(dirs):
if success:
nodes += [node]
else:
util.output("cannot create working directory for %s" % node.name)
results += [(node, False)]
# Start Bro process.
cmds = []
envs = []
for node in nodes:
cmds += [(node, "start", [node.cwd()] + _makeBroParams(node, True))]
envs += [_makeEnvParam(node)]
nodes = []
for (node, success, output) in execute.runHelperParallel(cmds, envs=envs):
if success:
nodes += [node]
node.setPID(int(output[0]))
else:
util.output("cannot start %s" % node.name)
results += [(node, False)]
# Check whether processes did indeed start up.
hanging = []
running = []
for (node, success) in waitForBros(nodes, "RUNNING", 3, True):
if success:
running += [node]
else:
hanging += [node]
# It can happen that Bro hangs in DNS lookups at startup
# which can take a while. At this point we already know
# that the process has been started (waitForBro ensures that).
# If by now there is not a TERMINATED status, we assume that it
# is doing fine and will move on to RUNNING once DNS is done.
for (node, success) in waitForBros(hanging, "TERMINATED", 0, False):
if success:
util.output("%s terminated immediately after starting; check output with \"diag\"" % node.name)
node.clearPID()
results += [(node, False)]
else:
util.output("(%s still initializing)" % node.name)
running += [node]
for node in running:
cron.logAction(node, "started")
results += [(node, True)]
return results
if n == manager:
continue
if not execute.isAlive(n.addr):
continue
nodes += [n]
if config.Config.havenfs != "1":
# Non-NFS, need to explicitly synchronize.
dirs = []
for dir in [config.Config.subst(dir) for (dir, mirror) in Syncs if not mirror]:
dirs += [(n, dir) for n in nodes]
for (node, success) in execute.mkdirs(dirs):
if not success:
util.warn("cannot create directory %s on %s" % (dir, node.name))
paths = [config.Config.subst(dir) for (dir, mirror) in Syncs if mirror]
execute.sync(nodes, paths)
util.output("done.")
# Note: the old code created $brobase explicitly but it seems the loop above should
# already take care of that.
else:
# NFS. We only need to take care of the spool/log directories.
paths = [config.Config.spooldir]
paths += [config.Config.tmpdir]
if not execute.isAlive(n.addr):
continue
nodes += [n]
if config.Config.havenfs != "1":
# Non-NFS, need to explicitly synchronize.
dirs = []
for dir in [
config.Config.subst(dir) for (dir, mirror) in Syncs
if not mirror
]:
dirs += [(n, dir) for n in nodes]
for (node, success) in execute.mkdirs(dirs):
if not success:
util.warn("cannot create directory %s on %s" %
(dir, node.name))
paths = [config.Config.subst(dir) for (dir, mirror) in Syncs if mirror]
execute.sync(nodes, paths)
util.output("done.")
# Note: the old code created $brobase explicitly but it seems the loop above should
# already take care of that.
else:
# NFS. We only need to take care of the spool/log directories.
paths = [config.Config.spooldir]
paths += [config.Config.tmpdir]
def _startNodes(nodes):
results = []
filtered = []
# Ignore nodes which are still running.
for (node, isrunning) in isRunning(nodes):
if not isrunning:
filtered += [node]
util.output("starting %s ..." % node.name)
else:
util.output("%s still running" % node.name)
nodes = filtered
# Generate crash report for any crashed nodes.
crashed = [node for node in nodes if node.hasCrashed()]
_makeCrashReports(crashed)
# Make working directories.
dirs = [(node, node.cwd()) for node in nodes]
nodes = []
for (node, success) in execute.mkdirs(dirs):
if success:
nodes += [node]
else:
util.output("cannot create working directory for %s" % node.name)
results += [(node, False)]
# Start Bro process.
cmds = []
envs = []
for node in nodes:
cmds += [(node, "start", [node.cwd()] + _makeBroParams(node, True))]
envs += [_makeEnvParam(node)]
nodes = []
for (node, success, output) in execute.runHelperParallel(cmds, envs=envs):
if success:
nodes += [node]
node.setPID(int(output[0]))
else:
util.output("cannot start %s" % node.name)
results += [(node, False)]
# Check whether processes did indeed start up.
hanging = []
running = []
for (node, success) in waitForBros(nodes, "RUNNING", 3, True):
if success:
running += [node]
else:
hanging += [node]
# It can happen that Bro hangs in DNS lookups at startup
# which can take a while. At this point we already know
# that the process has been started (waitForBro ensures that).
# If by now there is not a TERMINATED status, we assume that it
# is doing fine and will move on to RUNNING once DNS is done.
for (node, success) in waitForBros(hanging, "TERMINATED", 0, False):
if success:
util.output(
"%s terminated immediately after starting; check output with \"diag\""
% node.name)
node.clearPID()
results += [(node, False)]
else:
util.output("(%s still initializing)" % node.name)
running += [node]
for node in running:
cron.logAction(node, "started")
results += [(node, True)]
return results
