linse/frappy
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

improve error handling

Browse Source 
in the current implementation, in case polling results in an error,
an error_update is generated. But on activate, for this parameter
a normal update will be generated, indicating that this value is ok.

With this change, parameters have an additonal attribue 'readerror',
which save the error generated on the last read, or None, if the
read was successful.

In addition, subsequent repeated errors are only reported the first
time.

Change-Id: I39f18d17debadd8aa5b904e59998d2172f3f11b9
Reviewed-on: https://forge.frm2.tum.de/review/c/sine2020/secop/playground/+/21974
Tested-by: JenkinsCodeReview <bjoern_pedersen@frm2.tum.de>
Reviewed-by: Markus Zolliker <markus.zolliker@psi.ch>
This commit is contained in:
zolliker
2019-12-06 15:14:16 +01:00
parent 94a539da90
commit 44eeea1159
7 changed files with 105 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
secop/poller.py
View File

@ -36,7 +36,7 @@ Usage examples:
import time
from threading import Event
from heapq import heapify, heapreplace
from secop.lib import mkthread, formatException
from secop.lib import mkthread
from secop.errors import ProgrammingError
# poll types:
@ -137,8 +137,7 @@ class Poller(PollerBase):
# later, they will be converted to heaps
for pname, pobj in module.parameters.items():
polltype = pobj.poll
rfunc = getattr(module, 'read_' + pname, None)
if not polltype or not rfunc:
if not polltype:
continue
if not hasattr(module, 'pollinterval'):
raise ProgrammingError("module %s must have a pollinterval"
@ -159,7 +158,7 @@ class Poller(PollerBase):
handlers.add(pobj.handler)
# placeholders 0 are used for due, lastdue and idx
self.queues[polltype].append((0, 0,
(0, module, pobj, rfunc, factors[polltype])))
(0, module, pobj, pname, factors[polltype])))
def poll_next(self, polltype):
'''try to poll next item
@ -177,7 +176,7 @@ class Poller(PollerBase):
due, lastdue, pollitem = queue[0]
if now < due:
return due
_, module, pobj, rfunc, factor = pollitem
_, module, pobj, pname, factor = pollitem
if polltype == DYNAMIC and not module.isBusy():
interval = module.pollinterval # effective interval
@ -187,11 +186,7 @@ class Poller(PollerBase):
mininterval = interval
due = max(lastdue + interval, pobj.timestamp + interval * 0.5)
if now >= due:
try:
rfunc()
except Exception: # really all. errors are handled within rfunc
# TODO: filter repeated errors and log just statistics
module.log.error(formatException())
module.pollOne(pname)
done = True
lastdue = due
due = max(lastdue + mininterval, now + min(self.maxwait, mininterval * 0.5))
@ -217,16 +212,13 @@ class Poller(PollerBase):
return
# do all polls once and, at the same time, insert due info
for _, queue in sorted(self.queues.items()): # do SLOW polls first
for idx, (_, _, (_, module, pobj, rfunc, factor)) in enumerate(queue):
for idx, (_, _, (_, module, pobj, pname, factor)) in enumerate(queue):
lastdue = time.time()
try:
rfunc()
except Exception: # really all. errors are handled within rfunc
module.log.error(formatException())
module.pollOne(pname)
due = lastdue + min(self.maxwait, module.pollinterval * factor)
# in python 3 comparing tuples need some care, as not all objects
# are comparable. Inserting a unique idx solves the problem.
queue[idx] = (due, lastdue, (idx, module, pobj, rfunc, factor))
queue[idx] = (due, lastdue, (idx, module, pobj, pname, factor))
heapify(queue)
started_callback() # signal end of startup
nregular = len(self.queues[REGULAR])