Hi Team, Need help on the stack trace still appears in the logs though the flag is enabled and the system is stable but worried about the stack trace in the logs.
---------- Forwarded message --------- From: Raghunath Mahakud <mahakudraghunath...@gmail.com> Date: Thu, Mar 20, 2025 at 2:25 PM Subject: Re: OOM issue with ruta script To: Peter Klügl <pklu...@gmail.com> what is the impact of enabling the flag could be less rules match and less entity detection On Wed, Mar 19, 2025 at 4:57 PM Raghunath Mahakud < mahakudraghunath...@gmail.com> wrote: > Hi Peter, > > any thoughts on exception stack trace though system is stable and working > > On Wed, Mar 19, 2025 at 9:32 AM Raghunath Mahakud < > mahakudraghunath...@gmail.com> wrote: > >> https://uima.apache.org/d/ruta-current/ruta.html >> >> i didn't find any section for rewriting the rules. >> >> On Tue, Mar 18, 2025 at 8:34 PM Raghunath Mahakud < >> mahakudraghunath...@gmail.com> wrote: >> >>> Hi Peter, >>> >>> Sorry to bug you always but can you check once why error logs and what >>> is happening though the system is stable now? >>> >>> Is any optimization needed for the ruta script and the flag activation >>> is not enough any links for that? >>> >>> >>> >>> >>> On Tue, Mar 18, 2025 at 7:58 PM Raghunath Mahakud < >>> mahakudraghunath...@gmail.com> wrote: >>> >>>> it's working after activating SimpleGreedyForComposed entities getting >>>> extracted and the system is stable. >>>> >>>> But in the logs it's showing stackoverflow error, is the above config >>>> waits for error logs? >>>> >>>> but without this fix system went down and nothing works. >>>> >>>> Failed to run script: address >>>> java.lang.StackOverflowError: null >>>> at java.util.TreeMap.addEntryToEmptyMap(TreeMap.java:776) ~[?:?] >>>> at java.util.TreeMap.put(TreeMap.java:785) ~[?:?] >>>> at java.util.TreeMap.put(TreeMap.java:534) ~[?:?] >>>> at java.util.TreeSet.add(TreeSet.java:255) ~[?:?] >>>> at java.util.AbstractCollection.addAll(AbstractCollection.java:336) >>>> ~[?:?] >>>> at java.util.TreeSet.addAll(TreeSet.java:309) ~[?:?] >>>> at >>>> org.apache.uima.ruta.rule.ComposedRuleElementMatch.getTextsMatched(ComposedRuleElementMatch.java:285) >>>> ~[ruta-core-3.3.0.jar:?] >>>> at >>>> org.apache.uima.ruta.rule.ComposedRuleElement.doMatch(ComposedRuleElement.java:556) >>>> ~[ruta-core-3.3.0.jar:?] >>>> at >>>> org.apache.uima.ruta.rule.ComposedRuleElement.fallbackContinue(ComposedRuleElement.java:421) >>>> ~[ruta-core-3.3.0.jar:?] >>>> at >>>> org.apache.uima.ruta.rule.RutaRuleElement.continueMatchSomewhereElse(RutaRuleElement.java:211) >>>> ~[ruta-core-3.3.0.jar:?] >>>> >>>> >>>> On Tue, Mar 18, 2025 at 2:46 PM Raghunath Mahakud < >>>> mahakudraghunath...@gmail.com> wrote: >>>> >>>>> See my answer to the last topic how to rewrite the rules. >>>>> Can you share the link please? >>>>> >>>>> On Tue, Mar 18, 2025 at 2:36 PM Peter Klügl <pklu...@gmail.com> wrote: >>>>> >>>>>> It changes the inference to an algorithm that does not use recursion. >>>>>> >>>>>> >>>>>> About disjunctive and quantifiers: do not use "|" and no "*" or "+" >>>>>> after parentheses. >>>>>> >>>>>> See my answer to the last topic how to rewrite the rules. >>>>>> >>>>>> >>>>>> Best >>>>>> >>>>>> >>>>>> Peter >>>>>> >>>>>> >>>>>> >>>>>> Am 18.03.2025 um 09:50 schrieb Raghunath Mahakud: >>>>>> >>>>>> ``` >>>>>> stream.setDynamicAnchoring(dynamicAnchoring); >>>>>> stream.setGreedyRuleElement(false); >>>>>> stream.setGreedyRule(false); >>>>>> stream.setSimpleGreedyForComposed(true); >>>>>> ``` >>>>>> now i added the new flag setSimpleGreedyForComposed but how this >>>>>> going to help? >>>>>> >>>>>> On Tue, Mar 18, 2025 at 2:11 PM Raghunath Mahakud < >>>>>> mahakudraghunath...@gmail.com> wrote: >>>>>> >>>>>>> oho got now new flag setSimpleGreedyForComposed you mean we need to >>>>>>> enable this right. >>>>>>> but can you help in checking the ruta script once and the >>>>>>> disjunctive composed rule elements part >>>>>>> >>>>>>> On Tue, Mar 18, 2025 at 2:08 PM Raghunath Mahakud < >>>>>>> mahakudraghunath...@gmail.com> wrote: >>>>>>> >>>>>>>> *setGreedyRule is disabled* for us and can you see the script once >>>>>>>> and let me know if we have any disjunctive composed rule elements and >>>>>>>> especially quantifiers at composed elements. >>>>>>>> I didn't understand this part of disjunctive composed rule >>>>>>>> elements and especially quantifiers at composed elements. >>>>>>>> >>>>>>>> On Tue, Mar 18, 2025 at 2:05 PM Peter Klügl <pklu...@gmail.com> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> Hi, >>>>>>>>> >>>>>>>>> >>>>>>>>> you should rewrite your rules and avoid the disjunctive composed >>>>>>>>> rule elements and especially quantifiers at composed elements. >>>>>>>>> >>>>>>>>> The inference for those rules can "explode" leading to OOM and >>>>>>>>> similar problems. >>>>>>>>> >>>>>>>>> >>>>>>>>> There is also a parameter "simpleGreedyForComposed" which could >>>>>>>>> try to activate. >>>>>>>>> >>>>>>>>> >>>>>>>>> Best >>>>>>>>> >>>>>>>>> >>>>>>>>> Peter >>>>>>>>> >>>>>>>>> >>>>>>>>> Am 18.03.2025 um 09:27 schrieb Raghunath Mahakud: >>>>>>>>> >>>>>>>>> Hi Peter, >>>>>>>>> >>>>>>>>> attached is the ruta script and texts we are testing with and we >>>>>>>>> have setted >>>>>>>>> PARAM_MAX_RULE_MATCHES and PARAM_MAX_RULE_ELEMENT_MATCHES to 12000 >>>>>>>>> but still getting OOM. >>>>>>>>> >>>>>>>>> When testing with vanilla ruta workbench there also the issue is >>>>>>>>> reproducible. >>>>>>>>> >>>>>>>>> [image: image.png] >>>>>>>>> >>>>>>>>>
