just started getting this IOError: write: broken pipe (EPIPE)

[lbilli]

just started getting this IOError: write: broken pipe (EPIPE)

[lbilli]

It looks you are getting disconnected by TWS, probably due to pacing issues.
TWS doesn't allow more then 50 requests per second, over any 5 second periods; i.e. 250 req / 5s.

You should check if you are getting any of these error messages:

  0 Warning: Approaching max rate of 50 messages per second (48)
 or
100 Max rate of messages per second has been exceeded:max=50

After sending error 100 TWS closes the connection immediately.

Workarounds:

• carefully pace requests yourself not to exceed the threshold of 250 req / 5s
• TWS before 10.16: the default behavior is to disconnect when threshold is breached.
  Use connectOptions="+PACEAPI" when connecting and TWS will automatically pace the requests and avoid disconnections.
• TWS 10.16 and later: the default behavior is to automatically pace requests unless option
  API -> "Reject messages above maximum allowed message rate vs. applying pacing." is checked.

Ref: IB Release Notes

If running TWS pre 10.16 you might want to consider upgrading to a current version and keep the default behavior.

It looks you are getting disconnected by TWS, probably due to pacing issues. TWS doesn't allow more then 50 requests per second, over any 5 second periods; i.e. 250 req / 5s.

You should check if you are getting any of these error messages:

  0 Warning: Approaching max rate of 50 messages per second (48)
 or
100 Max rate of messages per second has been exceeded:max=50

After sending error 100 TWS closes the connection immediately.

Workarounds:

* carefully pace requests yourself not to exceed the threshold of 250 req / 5s

* TWS before 10.16: the default behavior is to disconnect when threshold is breached.
  Use `connectOptions="+PACEAPI"` when connecting and TWS will automatically pace the requests and avoid disconnections.

* TWS 10.16 and later: the default behavior is to automatically pace requests unless option
  API -> "Reject messages above maximum allowed message rate vs. applying pacing." is checked.

Ref: IB Release Notes

If running TWS pre 10.16 you might want to consider upgrading to a current version and keep the default behavior.

thank you so much for taking the time.

this code has been working for about 3 months without fault so this behavior started 12/16/22.

I checked the HELP for the TWS desktop I am using as a gateway and the build is 10.20.1f 12/13/22

I don't know how to put connectOptions="+PACEAPI" into my code.

ALSO I can't figure out how to check the log files. I "thought" that the log file was given in launcher.log but now all it says

2022-12-20 21:00:21.540 [SA] INFO  [JTS-PostAuthenticateS2-19] - performLogSwitch() [force=false]
2022-12-20 21:00:21.657 [SA] INFO  [JTS-PostAuthenticateS2-19] - Switching to a new encrypted log..

[lbilli]

I don't know how to put connectOptions="+PACEAPI" into my code.

That's an argument of Jib.connect(), as in:

ib = Jib.connect(4002, 1, "+PACEAPI")

Though, it shouldn't be necessary for TWS 10.20 if
Settings -> API -> "Reject messages above maximum allowed message rate vs. applying pacing." is unchecked.

You should check if you are getting any of these error messages:

 0 Warning: Approaching max rate of 50 messages per second (48)
or
100 Max rate of messages per second has been exceeded:max=50

After sending error 100 TWS closes the connection immediately.

ALSO I can't figure out how to check the log files. I "thought" that the log file was given in launcher.log but now all it says

These messages are not found in launcher.log. TWS sends them directly to your program, which should handle them properly via the callback:

wrap.error(id, errorCode, errorString, advancedOrderRejectJson)

Your wrap.error() should at least print or save the messages somewhere for diagnostic purposes.

I don't know how to put connectOptions="+PACEAPI" into my code.

That's an argument of Jib.connect(), as in:

ib = Jib.connect(4002, 1, "+PACEAPI")

Though, it shouldn't be necessary for TWS 10.20 if Settings -> API -> "Reject messages above maximum allowed message rate vs. applying pacing." is unchecked.

You should check if you are getting any of these error messages:

 0 Warning: Approaching max rate of 50 messages per second (48)
or
100 Max rate of messages per second has been exceeded:max=50

After sending error 100 TWS closes the connection immediately.

ALSO I can't figure out how to check the log files. I "thought" that the log file was given in launcher.log but now all it says

These messages are not found in launcher.log. TWS sends them directly to your program, which should handle them properly via the callback:

wrap.error(id, errorCode, errorString, advancedOrderRejectJson)

Your wrap.error() should at least print or save the messages somewhere for diagnostic purposes.

thank you for the quick reply.

I can confirm that Settings -> API -> "Reject messages above maximum allowed message rate vs. applying pacing." is unchecked.

I have the code from your EXCELLENT example (see below) is this enough to diagnose this issue please?
I could add logging to the code ( which I should have done in the first place).

wrap = Jib.Wrapper(
         # These events are always sent
         error= (id, errorCode, errorString, advancedOrderRejectJson) -> println("DJ Error: $(something(id, "NA")) $errorCode $errorString $advancedOrderRejectJson"),

REPLACED BY

wrap = Jib.Wrapper(
         # These events are always sent
         # error= (id, errorCode, errorString, advancedOrderRejectJson) -> println("DJ Error: $(something(id, "NA")) $errorCode $errorString $advancedOrderRejectJson"),
         error= (id, errorCode, errorString, advancedOrderRejectJson) ->  @info "message : $(now()) : DJ Error: $(something(id, "NA")) $errorCode $errorString $advancedOrderRejectJson"   ),

Got to say that it's weird this JUST started happening. I have a raspberry pi monitoring the ISP feed and there are no problems there.

[lbilli]

That is the wrap.error() I was referring to.

If the error messages I mentioned above don't show up in your logs then the root problem might be something different than sending too many requests too quickly.

If pacing is not the issue, browsing through your logs might give some clues of what is going on.

You can verify that your application indeed gets unexpectedly disconnected from TWS by checking:

isopen(ib.socket)  # = true when connected 

or by clicking on DATA on the upper right corner of the TWS window.

problem just went away this morning BUT I am certainly going to go back and do this properly using your advice. Thank you for a wonderful development environment. I can't spend more than a hour a week on this BUT in that hour your code makes a HUGE difference. I learn by example and you REALLY do a great job. Thanks and happy hols to you and yours