Hi Chris Thanks for your reply. My work around is in fact, that I set the scheduling to 1 sec. And hopefully it is fixed and well tested to the next release. I just can't understand why a bug like this should come in a new release.
Kind regards Jens M. Kofoed Den tir. 1. feb. 2022 kl. 12.24 skrev Chris Sampson < chris.samp...@naimuri.com>: > Jens, > > A quick look at the first Jira ticket you linked (NIFI-9564) mentions that > the processors have the "@TriggerWhenEmpty" annotation specified, which > means that they will continually trigger according to the configured > Scheduling even when there are no incoming FlowFiles. Whatsmore is that the > processors don't actually do anything when triggered with no input FlowFile > - so it's a completely wasted use of resources (N.B. I'm not really > familiar with these processors, just reading the Jira ticket description > and taking a very quick look at the linked PR). > > They also don't currently include the "@SupportsBatching" annotation, > which can drastically improve performance when processing multiple input > FlowFiles in a lot of processors. > > This Jira ticket has been resolved for 1.16.0, so hopefully the next > feature release of NiFi will include the improvements. Until then, you > could reduce the Scheduling of the processor to at least have it run fewer > tasks (e.g. process every 1 second instead of 0 seconds). This would reduce > the number of tasks/time, but also mean that any incoming FlowFiles would > have to wait up to 1 second before being taken from the queue for > processing, so you have to decide whether it's more important to cut down > on resource waste or have the reduced latency for incoming FlowFiles. I've > faced similar considerations for things like GetSQS in the past - I either > leave it at "0 secs" and let it run continually or change it to "1 sec" and > accept that some messages landing on the SQS queue will take a little > longer to be picked up by NiFi. > > > > --- > *Chris Sampson* > IT Consultant > chris.samp...@naimuri.com > > > On Tue, 1 Feb 2022 at 10:42, Jens M. Kofoed <jmkofoed....@gmail.com> > wrote: > >> Hi >> >> I don't know if these JIRAs are handling the same error I got with NIFI >> v. 1.15.3 >> https://issues.apache.org/jira/browse/NIFI-9546 PutTCP / PutUDP are >> inefficient >> https://issues.apache.org/jira/browse/NIFI-9571 PutTCP and PutUDP not >> committing Session >> >> But the PutCP process runs tasks continuously without any incoming flow >> files. >> Here the process has only being running for a couple of seconds and >> already have run 3,721,047 tasks. >> [image: image.png] >> >> kind regards >> Jens M. Kofoed >> >>
