import sys
import string
import time
-import soltesz
-import plc
-import comon
+from datetime import datetime,timedelta
import threadpool
+import threading
-from config import config
-from optparse import OptionParser
-parser = OptionParser()
-parser.set_defaults(filename="", increment=False, dbname="findbadnodes")
-parser.add_option("-f", "--nodes", dest="filename", metavar="FILE",
- help="Provide the input file for the node list")
-parser.add_option("", "--dbname", dest="dbname", metavar="FILE",
- help="Specify the name of the database to which the information is saved")
-parser.add_option("-i", "--increment", action="store_true", dest="increment",
- help="Increment round number to force refresh or retry")
-config = config(parser)
-config.parse_args()
+from monitor.util import file
+from pcucontrol.util import command
+from monitor import config
+from monitor.database.info.model import FindbadNodeRecord, session
+
+from monitor.sources import comon
+from monitor.wrapper import plc, plccache
+from monitor.scanapi import *
+
+from nodequery import verify,query_to_dict,node_select
+import traceback
+from monitor.common import nmap_port_status
+
+#print "starting sqlfindbad.py"
# QUERY all nodes.
COMON_COTOPURL= "http://summer.cs.princeton.edu/status/tabulator.cgi?" + \
- "table=table_nodeview&" + \
- "dumpcols='name,resptime,sshstatus,uptime,lastcotop'&" + \
- "formatcsv"
+ "table=table_nodeview&" + \
+ "dumpcols='name,resptime,sshstatus,uptime,lastcotop,cpuspeed,memsize,disksize'&" + \
+ "formatcsv"
#"formatcsv&" + \
#"select='lastcotop!=0'"
+api = plc.getAuthAPI()
+plc_lock = threading.Lock()
round = 1
-externalState = {'round': round, 'nodes': {}}
+global_round = round
count = 0
-def collectPingAndSSH(nodename, cohash):
- ### RUN PING ######################
- ping = soltesz.CMD()
- (oval,eval) = ping.run_noexcept("ping -c 1 -q %s | grep rtt" % nodename)
-
- values = {}
-
- if oval == "":
- # An error occurred
- values['ping'] = "NOPING"
- else:
- values['ping'] = "PING"
-
- ### RUN SSH ######################
- b_getbootcd_id = True
- ssh = soltesz.SSH('root', nodename)
- oval = ""
- eval = ""
- (oval, eval) = ssh.run_noexcept('echo `uname -a ; ls /tmp/bm.log`')
- val = oval
- if "2.6.17" in oval or "2.6.20" in oval:
- values['ssh'] = 'SSH'
- if "bm.log" in oval:
- values['category'] = 'ALPHA'
- values['state'] = 'DEBUG'
- else:
- values['category'] = 'ALPHA'
- values['state'] = 'BOOT'
- elif "2.6.12" in oval or "2.6.10" in oval:
- values['ssh'] = 'SSH'
- values['category'] = 'PROD'
- if "bm.log" in oval:
- values['state'] = 'DEBUG'
- else:
- values['state'] = 'BOOT'
- elif "2.4" in oval:
- b_getbootcd_id = False
- values['ssh'] = 'SSH'
- values['category'] = 'OLDBOOTCD'
- values['state'] = 'DEBUG'
- elif oval != "":
- values['ssh'] = 'SSH'
- values['category'] = 'UNKNOWN'
- if "bm.log" in oval:
- values['state'] = 'DEBUG'
- else:
- values['state'] = 'BOOT'
- else:
- # An error occurred.
- b_getbootcd_id = False
- values['ssh'] = 'NOSSH'
- values['category'] = 'ERROR'
- values['state'] = 'DOWN'
- val = eval.strip()
-
- values['kernel'] = val
-
- if b_getbootcd_id:
- # try to get BootCD for all nodes that are not 2.4 nor inaccessible
- (oval, eval) = ssh.run_noexcept('cat /mnt/cdrom/bootme/ID')
- val = oval
- if "BootCD" in val:
- values['bootcd'] = val
- if "v2" in val:
- values['category'] = 'OLDBOOTCD'
- else:
- values['bootcd'] = ""
- else:
- values['bootcd'] = ""
-
- # TODO: get bm.log for debug nodes.
- # 'zcat /tmp/bm.log'
-
- values['comonstats'] = cohash[nodename]
- # include output value
- ### GET PLC NODE ######################
- d_node = plc.getNodes({'hostname': nodename})
- site_id = -1
- if d_node and len(d_node) > 0:
- pcu = d_node[0]['pcu_ids']
- if len(pcu) > 0:
- values['pcu'] = "PCU"
- else:
- values['pcu'] = "NOPCU"
- site_id = d_node[0]['site_id']
- values['plcnode'] = {'status' : 'SUCCESS', 'pcu_ids': pcu, 'site_id': site_id}
- else:
- values['pcu'] = "UNKNOWN"
- values['plcnode'] = {'status' : "GN_FAILED"}
-
-
- ### GET PLC SITE ######################
- d_site = plc.getSites({'site_id': site_id})
- if d_site and len(d_site) > 0:
- max_slices = d_site[0]['max_slices']
- num_slices = len(d_site[0]['slice_ids'])
- num_nodes = len(d_site[0]['node_ids'])
- loginbase = d_site[0]['login_base']
- values['plcsite'] = {'num_nodes' : num_nodes,
- 'max_slices' : max_slices,
- 'num_slices' : num_slices,
- 'login_base' : loginbase,
- 'status' : 'SUCCESS'}
- else:
- values['plcsite'] = {'status' : "GS_FAILED"}
-
- return (nodename, values)
-
-def recordPingAndSSH(request, result):
- global externalState
- global count
- (nodename, values) = result
-
- global_round = externalState['round']
- externalState['nodes'][nodename]['values'] = values
- externalState['nodes'][nodename]['round'] = global_round
-
- count += 1
- print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
- soltesz.dbDump(config.dbname, externalState)
-
# this will be called when an exception occurs within a thread
def handle_exception(request, result):
print "Exception occured in request %s" % request.requestID
for i in result:
print "Result: %s" % i
-
+
def checkAndRecordState(l_nodes, cohash):
- global externalState
+ global global_round
global count
- global_round = externalState['round']
tp = threadpool.ThreadPool(20)
+ scannode = ScanNodeInternal(global_round)
# CREATE all the work requests
for nodename in l_nodes:
- if nodename not in externalState['nodes']:
- externalState['nodes'][nodename] = {'round': 0, 'values': []}
+ #fbnodesync = FindbadNodeRecordSync.findby_or_create(hostname=nodename, if_new_set={'round':0})
+ #node_round = fbnodesync.round
+ node_round = global_round - 1
+ #fbnodesync.flush()
- node_round = externalState['nodes'][nodename]['round']
- if node_round < global_round:
+ if node_round < global_round or config.force:
# recreate node stats when refreshed
#print "%s" % nodename
- req = threadpool.WorkRequest(collectPingAndSSH, [nodename, cohash], {},
- None, recordPingAndSSH, handle_exception)
+ req = threadpool.WorkRequest(scannode.collectInternal, [nodename, cohash], {},
+ None, scannode.record, handle_exception)
tp.putRequest(req)
else:
# We just skip it, since it's "up to date"
count += 1
- print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
- pass
+ #print "%d %s %s" % (count, nodename, externalState['nodes'][nodename]['values'])
+ print "%d %s %s" % (count, nodename, node_round)
# WAIT while all the work requests are processed.
+ begin = time.time()
while 1:
try:
time.sleep(1)
tp.poll()
+ # if more than two hours
+ if time.time() - begin > (60*60*1.5):
+ print "findbad.py has run out of time!!!!!!"
+ os._exit(1)
except KeyboardInterrupt:
print "Interrupted!"
break
print "All results collected."
break
-
+ #print FindbadNodeRecordSync.query.count()
+ print FindbadNodeRecord.query.count()
+ session.flush()
def main():
- global externalState
+ global global_round
- externalState = soltesz.if_cached_else(1, config.dbname, lambda : externalState)
+ #fbsync = FindbadNodeRecordSync.findby_or_create(hostname="global",
+ # if_new_set={'round' : global_round})
+ #global_round = fbsync.round
if config.increment:
# update global round number to force refreshes across all nodes
- externalState['round'] += 1
+ global_round += 1
cotop = comon.Comon()
# lastcotop measures whether cotop is actually running. this is a better
# metric than sshstatus, or other values from CoMon
cotop_url = COMON_COTOPURL
- cohash = cotop.coget(cotop_url)
+ # history information for all nodes
+ cohash = {}
+ #cohash = cotop.coget(cotop_url)
+ l_nodes = plccache.l_nodes
+ if config.nodelist:
+ f_nodes = file.getListFromFile(config.nodelist)
+ l_nodes = filter(lambda x: x['hostname'] in f_nodes, l_nodes)
+ elif config.node:
+ f_nodes = [config.node]
+ l_nodes = filter(lambda x: x['hostname'] in f_nodes, l_nodes)
+ elif config.nodegroup:
+ ng = api.GetNodeGroups({'name' : config.nodegroup})
+ l_nodes = plccache.GetNodesByIds(ng[0]['node_ids'])
+ elif config.site:
+ site = plccache.GetSitesByName([config.site])
+ l_nodes = plccache.GetNodesByIds(site[0]['node_ids'])
+ elif config.sitelist:
+ site_list = config.sitelist.split(',')
+ sites = plccache.GetSitesByName(site_list)
+ node_ids = []
+ for s in sites:
+ node_ids += s['node_ids']
+ l_nodes = plccache.GetNodesByIds(node_ids)
+
+ l_nodes = [node['hostname'] for node in l_nodes]
+
+ # perform this query after the above options, so that the filter above
+ # does not break.
+ if config.nodeselect:
+ plcnodes = plccache.l_nodes
+ plcnodes = [ node['hostname'] for node in plcnodes ]
+ l_nodes = node_select(config.nodeselect, plcnodes, None)
- if config.filename == "":
- l_nodes = cohash.keys()
- else:
- l_nodes = config.getListFromFile(config.filename)
+ print "fetching %s hosts" % len(l_nodes)
checkAndRecordState(l_nodes, cohash)
+ if config.increment:
+ # update global round number to force refreshes across all nodes
+ #fbsync.round = global_round
+ #fbsync.flush()
+ pass
+
return 0
if __name__ == '__main__':
+ from monitor import parser as parsermodule
+
+ parser = parsermodule.getParser(['nodesets'])
+
+ parser.set_defaults( increment=False, dbname="findbad", cachenodes=False,
+ force=False,)
+ parser.add_option("", "--cachenodes", action="store_true",
+ help="Cache node lookup from PLC")
+ parser.add_option("", "--dbname", dest="dbname", metavar="FILE",
+ help="Specify the name of the database to which the information is saved")
+ parser.add_option("-i", "--increment", action="store_true", dest="increment",
+ help="Increment round number to force refresh or retry")
+ parser.add_option("", "--force", action="store_true", dest="force",
+ help="Force probe without incrementing global 'round'.")
+
+ parser = parsermodule.getParser(['defaults'], parser)
+
+ cfg = parsermodule.parse_args(parser)
+
try:
main()
except Exception, err:
+ print traceback.print_exc()
+ from monitor.common import email_exception
+ email_exception()
print "Exception: %s" % err
print "Saving data... exitting."
- soltesz.dbDump(config.dbname, externalState)
sys.exit(0)
+ print "sleeping"