def empty_return(self, dasquery, status='busy', reason=None):
"Return header/data when DAS server is busy"
if not reason:
reason = 'DAS server is busy'
reason += ', #requests=%s, #workers=%s, queue size=%s' \
% (self.reqmgr.size(), self.taskmgr.nworkers(), self.queue_limit)
head = dict(timestamp=time.time())
head.update({'status': status, 'reason': reason, 'ctime':0})
data = []
dasprint(dastimestamp('DAS INFO '), dasquery, 'server status=%s'%status, reason)
return self.datastream(dict(head=head, data=data))
def busy(self):
"""
Check server load and report busy status if
nrequests - nworkers > queue limit
"""
nrequests = self.reqmgr.size()
if (nrequests - self.taskmgr.nworkers()) > self.queue_limit:
msg = '#request=%s, queue_limit=%s, #workers=%s' \
% (nrequests, self.taskmgr.nworkers(), self.queue_limit)
dasprint(dastimestamp('DAS WEB SERVER IS BUSY '), msg)
return True
return False
def status(self):
"""Return list of all current requests in DAS queue"""
requests = [r for r in self.reqmgr.items()]
page = self.templatepage('das_status', requests=requests, time=time)
sdict = self.dasmgr.status()
sdict['web'] = self.taskmgr.status()
dasprint(dastimestamp('DAS INFO '), "web TaskManager", sdict['web'])
for key, val in sdict.items():
dasprint(dastimestamp('DAS INFO '), "%s TaskManager %s" % (key, val))
page += '<h3>Services</h3>'
def dump(idict):
"Dump input dict"
return ', '.join(['<em>%s:</em> %s' % (k, idict[k]) for k in sorted(idict)])
for key, val in sdict.items():
page += '<div>'
stats = ', '.join([dump(v) for v in val.values()])
page += '<b>%s</b>: %s' % (key, stats)
page += '</div>'
return self.page(page)
def check_pid(self, pid):
"""
Check status of given pid. This is a server callback
function for ajaxCheckPid, see js/ajax_utils.js
"""
# do not allow caching
set_no_cache_flags()
img = '<img src="%s/images/loading.gif" alt="loading"/>' % self.base
page = ''
try:
if self.taskmgr.is_alive(pid):
page = img + " processing PID=%s" % pid
else:
# at this point we don't know if request arrived to this host
# or it was processed. To distinguish the case we'll ask
# request manager for that pid
if self.reqmgr.has_pid(pid):
self.reqmgr.remove(pid)
self.taskmgr.remove(pid)
page = 'Request PID=%s is completed' % pid
page += ', please wait for results to load'
else:
# there're no request on this server, re-initiate it
ref = cherrypy.request.headers.get('Referer', None)
if ref:
url = urlparse(ref)
params = dict(parse_qsl(url.query))
return self.request(**params)
else:
msg = 'No referer in cherrypy.request.headers'
msg += '\nHeaders: %s' % cherrypy.request.headers
dasprint(dastimestamp('DAS WEB ERROR '), msg)
except Exception as err:
msg = 'check_pid fails for pid=%s' % pid
dasprint(dastimestamp('DAS WEB ERROR '), msg)
print_exc(err)
self.reqmgr.remove(pid)
self.taskmgr.remove(pid)
return self.error(gen_error_msg({'pid':pid}), wrap=False)
return page
def get_data(self, kwargs):
"""
Invoke DAS workflow and get data from the cache.
"""
head = dict(timestamp=time.time())
head['args'] = kwargs
uinput = kwargs.get('input', '')
inst = kwargs.get('instance', self.dbs_global)
idx = getarg(kwargs, 'idx', 0)
limit = getarg(kwargs, 'limit', 0) # do not impose limit
coll = kwargs.get('collection', 'merge')
status = kwargs.get('status')
error = kwargs.get('error')
reason = kwargs.get('reason')
dasquery = kwargs.get('dasquery', None)
time0 = time.time()
if dasquery:
dasquery = DASQuery(dasquery, instance=inst)
if dasquery.error:
return self.page(form + dasquery.error, ctime=time.time()-time0)
else:
check, content = \
self.generate_dasquery(uinput, inst, html_mode=False)
if check:
head.update({'status': 'fail', 'reason': content,
'ctime': time.time()-time0, 'input': uinput})
data = []
return head, data
dasquery = content # returned content is valid DAS query
try:
nres = self.dasmgr.nresults(dasquery, coll)
data = \
self.dasmgr.get_from_cache(dasquery, idx, limit)
# check that we got what we expected
data = [r for r in data]
if nres and not len(data):
for retry in range(1, 3, 5):
msg = 'retry in %s sec' % retry
dasprint(dastimestamp('DAS WARNING '), msg, dasquery)
time.sleep(retry) # retry one more time
data = \
self.dasmgr.get_from_cache(dasquery, idx, limit)
data = [r for r in data]
if len(data):
break
if nres and not len(data):
msg = 'fail to get all data for %s, nres=%s, len(data)=%s' \
% (dasquery, nres, len(data))
dasprint(dastimestamp('DAS WARNING '), msg)
status = 'fail'
reason = 'Fail to retrieve data from DAS cache, please retry'
if dasquery.aggregators:
# aggregators split DAS record into sub-system and then
# apply aggregator functions, therefore we need to correctly
# account for nresults. Resolve generator into list and take
# its length as nresults value.
data = [r for r in data]
nres = len(data)
if error: # DAS record contains an error
status = 'error'
head.update({'status':status, 'nresults':nres,
'ctime': time.time()-time0, 'dasquery': dasquery})
except Exception as exc:
status = 'fail'
reason = str(exc)
print_exc(exc)
head.update({'status': status,
'ctime': time.time()-time0, 'dasquery': dasquery})
data = []
head.update({'incache':self.dasmgr.incache(dasquery, coll='cache'),
'apilist':self.dasmgr.apilist(dasquery)})
if reason:
head.update({'reason': reason})
if status != 'ok':
head.update(self.info())
# check if query had dataset input and returned no results
# then run hint functions to find dataset in other DBS instances
mquery = dasquery.mongo_query
empty = False
for item in data:
if 'dataset.name' in mquery['spec'] and 'dataset' in mquery['fields'] \
and 'result' not in item:
if not item['dataset']:
empty = True
break
if empty: # if no results found add dataset from other DBS instances
hints = self.hint_datasets(kwargs)
for item in data:
item.update({'hints': hints})
return head, data