Hi Marton, I've just downloaded and compiled 0.10 minifi c++ The issue doesn't happen. When I stop the input port in NiFi I get this:
[org::apache::nifi::minifi::core::Processor] [warning] Caught "Process Session Operation: Can not find the transfer relationship for the updated flow 89289e16-4d90-11ec-bef4-c0b6f9f714b7" (N3org6apache4nifi6minifi9ExceptionE) during Processor::onTrigger of processor: 4d3e32ac-017d-1000-b4b1-e7f36e039202 (entrada) And when I start the input port again it comes back to work normally. The issue doesn't happen with MiNiFi 1.15 Java version too. I've created a Jira (https://issues.apache.org/jira/browse/MINIFICPP-1692) for this. Thanks. Regards. Eduardo Fontes On Wed, Nov 24, 2021 at 2:45 PM Marton Szasz <sza...@apache.org> wrote: > Hi Eduardo, > > It looks like the Remote Processor Group Port is blocking on some > call. I couldn't reproduce the issue with a similar setup, even after > stopping and restarting nifi multiple times. > > Could you create a Jira issue and attach your config files? A stack > trace could also be helpful, if you can start minifi in a debugger and > interrupt while it's stuck inside that onTrigger. > Can you reproduce the issue with the previous minifi c++ release? > > If you rely on site-to-site for some production workflow, I suggest > switching to MergeContent + InvokeHTTP and ListenHTTP on the nifi > side. This use case has received a lot of attention lately, so I think > it's both faster and more stable. > > Not sure if we should tank the release due to this issue, so I will > leave my vote in place. > > Thanks, > Marton > > On Wed, 24 Nov 2021 at 11:12, Eduardo Fontes <eduardo.fon...@gmail.com> > wrote: > > > > Hi all, > > > > I've tested minifi 0.11 + Nifi 1.15 and I got warnings that I don't know > if > > it's a version issue. > > > > How to reproduce: > > - Download and unpack Nifi 1.15 binary > > - Download, unpack and compile Minifi 0.11 source > > - Start Nifi with default settings, create a remote input port with name > > "entrada", start it and connect it to a funnel > > - Configure minifi TLS > > - Configure minifi's flow (config.yaml) with a ExecuteCommand (find ./ > each > > 30 secs) with success relationship to a RPG pointing to NiFi 1.15 > "entrada" > > input port > > - Start minifi > > > > At this point everything is ok. The result of the command "find ./" lands > > on the NiFi input port and stays in the funnel's queue. > > > > The problem is, if NiFi stops for some reason (a real world case), I > start > > to get warnings in minifi's log the like: > > [org::apache::nifi::minifi::SchedulingAgent] [warning] entrada::onTrigger > > has been running for 971026 ms in 4d3e32ac-017d-1000-b4b1-e7f36e039202 > > > > And if I start NiFi the warnings continue and it never stops. In this > > state, if I try to stop minifi using "minifi.sh stop" it doesn't stop > and I > > have to kill the minifi's process. > > It seems the thread stuck. > > > > Is this only with me? Am I doing something wrong? > > Please let me know. > > > > Thanks everybody. > > Eduardo Fontes > > > > On Wed, Nov 24, 2021 at 6:56 AM Pierre Villard < > pierre.villard...@gmail.com> > > wrote: > > > > > +1 (binding) > > > > > > Followed the usual steps to build the agent on my MBP and test it with > a > > > simple flow. > > > > > > Thanks Adam for taking care of this release! > > > > > > Pierre > > > > > > Le mer. 24 nov. 2021 à 10:49, Ádám Markovics <nullzer...@gmail.com> a > > > écrit : > > > > > > > +1 (non-binding) > > > > Verified and built on Ubuntu 20.04. Ran tests and also ran a simple > > > > (GetFile > > > > -PutFile) flow. > > > > Regards, > > > > > > > > Ádám > > > > > > > > On Mon, Nov 22, 2021 at 6:44 PM Marton Szasz <sza...@apache.org> > wrote: > > > > > > > > > +1 (binding) > > > > > > > > > > Followed the release helper guide. Tested compilation with both > GCC 11 > > > > > and Clang 13 (with libstdc++). > > > > > Used the convenience binary to collect systemd-journald messages > from > > > > > an ubuntu 16.04 VM and forward them via InvokeHTTP to a ListenHTTP > > > > > listener on host. The host was running the clang-compiled release > > > > > candidate agent. > > > > > Manually removed unused extensions now that we have them in > separate > > > > > .so files, just to see that it's still working. Everything went > > > > > smoothly. > > > > > > > > > > Thanks, > > > > > Marton > > > > > > > > > > > > > > > On Mon, 22 Nov 2021 at 16:08, Ferenc Gerlits <fgerl...@apache.org> > > > > wrote: > > > > > > > > > > > > +1 (non-binding) > > > > > > > > > > > > - verified hashes and signatures > > > > > > - compiled sources, ran unit tests > > > > > > - ran convenience binary with a Generate Flow File -> Log > Attribute > > > > flow > > > > > > using C2 with TLS > > > > > > > > > > > > Everything worked as expected. > > > > > > > > > > > > Thanks, > > > > > > Ferenc > > > > > > > > > > > > >
