These are just standard structures of the JVM. For example this is where things like the permgen live. The permgen is where in Java 7 and older class definitions and other things are loaded. We probably should just remove max for non-Heap as I'm not sure the idea of a max there makes sense. Now, regarding the fact that free space is such a small size compared to the total is something worth consideration. In the conf/bootstrap.conf you will see two really important lines which are commented out by default: #java.arg.11=-XX:PermSize=128M #java.arg.12=-XX:MaxPermSize=128M We do often recommend uncommenting those lines as the default size of the permgen is insufficient for how many classes NiFi can load at startup especially given its isolated classloader design. Simply uncommenting those lines will do the trick. Depending on the amount of classes that end up getting loaded and/or loading that may occur dynamically you may need more or less space allocated. If the permgen runs out of space entirely system performance will suffer so best to avoid this. We need to update the install guide to call this and all bootstrap properties out. We don't want to have these set by default because in Java 8 the permgen was restructured and is now part of the heap. Thanks! ... View more

Site-to-Site uses deflate at level 1 and compresses data in blocks/buffers. With site-to-site a series of 1..N flowfiles are sent at once and ack'd as a group. It is not configurable at this time. Keep in mind of course you can of course compress before sending to s2s and decompress after receiving from s2s using the CompressContent processor. Do you feel there would be a good bit of value in letting the compression of s2s be configurable? If so would that be for cases like where snappy makes sense because it is certain types of text data? Thanks Joe ... View more

The idea for making 'DOWNLOAD' its own event has been in play for a while.https://issues.apache.org/jira/browse/NIFI-10 We have sort of pushed this as we want to be really conservative making new event types. For REPLAY I think we call it a CLONE today (i think). If you'd like to revive that discussion and add the case for replay please go ahead and comment on that ticket. ... View more

There is no such alerting at this time. We do flag when expiration is set on connections and we've discussed extending that to include when backpressure is set but we don't have that now. We've also talked about actually signaling when expiration and/or backpressure are actually happening through the UI and of course the REST api but we've not done it yet. I think a good first approach would be a passive mechanism whereby we expose the fact of backpressure or expiration through the REST API so polling clients could know it (including our UI). ... View more

There are not any nightly builds in the Apache environment for NiFi to date. That is largely because it is really consistent with ASF policy even though you do see some other projects doing that. We could certainly setup something like that within HWX release engineering environment. ... View more

Puppet, Vagrant, and home grown goodness have all been used pretty extensively to ensure consistent configuration of servers and such. We need to keep doing more work though to separate concerns of 'box configuration' from 'nifi node' configuration. To the greatest extent possible NiFi's clustering should handle the 'nifi node' shared configuration needs and the box/os/security settings should be handled by systems like puppet or what the ops teams are most familiar with. ... View more