Random monitor timeout on read_attribute

[jkotan]

Hi,

From time to I get error

MainThread     INFO     2017-10-25 08:47:45,967 TaurusRootLogger: Using "PyQt4" for Qt
Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/nxsselector/Selector.py", line 941, in __restore
self.resetDescriptions()
  File "/usr/lib/python2.7/dist-packages/nxsselector/Selector.py", line 884, in resetDescriptions
self.storage.showErrors()
  File "/usr/lib/python2.7/dist-packages/nxsselector/Storage.py", line 276, in showErrors
errors = self.state.fetchErrors()
  File "/usr/lib/python2.7/dist-packages/nxsselector/ServerState.py", line 418, in fetchErrors
dc = self.__dp.read_attribute("descriptionErrors").value
  File "/usr/lib/python2.7/dist-packages/tango/green.py", line 264, in greener
ret = submitter(fn, self, *args, **kwargs)
  File "/usr/lib/python2.7/dist-packages/tango/green.py", line 100, in submit
return fn(*args, **kwargs)
  File "/usr/lib/python2.7/dist-packages/tango/device_proxy.py", line 352, in __DeviceProxy__read_attribute
return __check_read_attribute(self._read_attribute(value, extract_as))
PyTango.DevFailed: DevFailed[
DevError[
desc = Not able to acquire serialization (dev, class or process) monitor
  origin = TangoMonitor::get_monitor
  reason = API_CommandTimedOut
severity = ERR]

DevError[
desc = Failed to read_attribute on device p08/nxsrecselector/haspp08, attribute descriptionErrors
  origin = DeviceProxy::read_attribute()
  reason = API_AttributeFailed
severity = ERR]
]

when I read an attribute from tango server. It looks similar to #22

Is this issue was fixed? Is there any workaround?

I use pytango 9.2.1.

Best regards,
Jan

[jkotan]

To be more concrete DescriptionErrors is a tango attribute:

Writable     READ
Data format  Spectrum
Data type    DevString
Max Dim X    10000

of a tango server written in python.
Currently my workaround it a brutal force solution, i.e. a loop with read tries.

[vxgmichel]

Thanks for the report,

The issue you mentioned (#22) has been fixed in pytango 9.2.0, so that's not your problem here.

The thing is, the tango monitor lock timeout (Not able to acquire serialization monitor) can have many causes. It could be a deadlock, an overload, or any blocking code. You'll have to try to reproduce it with a minimalistic version of your server and share it with us so we can figure out what is going on.

[tiagocoutinho]

Hi @jkotan,

Are you running your server in any of the async modes (gevent/asyncio/...) ?

[jkotan]

cannot reproduce anymore