Jeremy Can you please file a JIRA to cover this scenario that you're observing. If you can share stack traces/thread dumps/logs. I agree this sounds like poor behavior and we should definitely make it better.
On Thu, Aug 24, 2017 at 2:04 PM, Jeremy Farbota <[email protected]> wrote: > I have a ListSFTP process that I've been using for a few months where a > partner posts some large files, I turn it on, then ListSFTP gets all the > filenames so a FetchSFTP can download them prior to other steps and > notification events later on. > > I had an issue this month initially because a password was changed. The > processor was unable to access the server (however it did not alert me to > that issue). Once I fixed the password, the primary node where this was > running still was not finding the new files. I switched to "All Nodes" and > one of the nodes was able to access the server. Unfortunately I had not yet > updated the password on the FetchSFTP processor so it failed to download. > When I fixed the password on that FetchSFTP (with 20 files still in the > queue), it would not run to download the files. > > Due to time constraints, I built another standalone service, moved the > template there, and it ran fine on that new, standalone NiFi on a different > server. > > I am able to SSH and SFTP into this SFTP server from the NiFi nodes in > question in the command line. I can list files from the command line as > well. I'm not being blacklisted by the host SFTP. > > I've restarted the NiFi nodes to see if that would help. It did not. > > I do not see the FetchSFTP service starting and stopping in nifi-app.log > when I turn it on and off. > > Another issue is that there is no failure or alert with the bad password. > If it fails, there is no response. Shouldn't there be an error if I put a > bad host or an "access denied" if I put in the wrong password? I ended up > figuring out the password was wrong by using a different client to figure > out why the NiFi processors were not responding. > > Any advice on how to get my ListSFTP and FetchSFTP responding again on my > main cluster? I'm hoping to continue using ListSFTP for other things on > prod but the processors are not able to be configured or ran at this time > on that cluster. When I try an create new processes, they do not respond. > > Thanks in advance! > > > [image: Payoff, Inc.] > *Jeremy Farbota* > Software Engineer, Data > Payoff, Inc. > > [email protected] > > >
