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

make startup faster in case of errors

Browse Source 
When the io of one SECoP module fails, it takes ages to startup
because each parameter poll takes the time to wait for a timeout.
After the first communication error on an io, no more startup polls
are tried on the modules using this io.

Change-Id: I0d250953dfe91a7d68d2d2b108395cc25d471afe
Reviewed-on: https://forge.frm2.tum.de/review/c/sine2020/secop/playground/+/28588
Tested-by: Jenkins Automated Tests <pedersen+jenkins@frm2.tum.de>
Reviewed-by: Enrico Faulhaber <enrico.faulhaber@frm2.tum.de>
Reviewed-by: Markus Zolliker <markus.zolliker@psi.ch>
This commit is contained in:
zolliker
2022-06-02 10:03:38 +02:00
parent e4aa2149f7
commit de5f17695c
2 changed files with 29 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
secop/modules.py
View File

@ -30,7 +30,7 @@ from functools import wraps
from secop.datatypes import ArrayOf, BoolType, EnumType, FloatRange, \
IntRange, StatusType, StringType, TextType, TupleOf, DiscouragedConversion
from secop.errors import BadValueError, ConfigError, \
from secop.errors import BadValueError, CommunicationFailedError, ConfigError, \
ProgrammingError, SECoPError, secop_error
from secop.lib import formatException, mkthread, UniqueObject, generalConfig
from secop.lib.enum import Enum
@ -641,7 +641,7 @@ class Module(HasAccessibles):
self.pollInfo.interval = fast_interval if flag else self.pollinterval
self.pollInfo.trigger()
def callPollFunc(self, rfunc):
def callPollFunc(self, rfunc, raise_com_failed=False):
"""call read method with proper error handling"""
try:
rfunc()
@ -658,6 +658,8 @@ class Module(HasAccessibles):
else:
# uncatched error: this is more serious
self.log.error('%s: %s', name, formatException())
if raise_com_failed and isinstance(e, CommunicationFailedError):
raise
def __pollThread(self, modules, started_callback):
"""poll thread body
@ -682,7 +684,7 @@ class Module(HasAccessibles):
trg.set()
self.registerReconnectCallback('trigger_polls', trigger_all)
# collect and call all read functions a first time
# collect all read functions
for mobj in modules:
pinfo = mobj.pollInfo = PollInfo(mobj.pollinterval, self.triggerPoll)
# trigger a poll interval change when self.pollinterval changes.
@ -693,7 +695,16 @@ class Module(HasAccessibles):
rfunc = getattr(mobj, 'read_' + pname)
if rfunc.poll:
pinfo.polled_parameters.append((mobj, rfunc, pobj))
mobj.callPollFunc(rfunc)
# call all read functions a first time
try:
for m in modules:
for mobj, rfunc, _ in m.pollInfo.polled_parameters:
mobj.callPollFunc(rfunc, raise_com_failed=True)
except CommunicationFailedError as e:
# when communication failed, probably all parameters and may be more modules are affected.
# as this would take a lot of time (summed up timeouts), we do not continue
# trying and let the server accept connections, further polls might success later
self.log.error('communication failure on startup: %s', e)
started_callback()
to_poll = ()
while True: