Re: FTP security question...

Hello Robert,

Am 11.04.2024 um 14:35 schrieb Robert Rogerson <rogersonra@xxxxxxxxx>:

Thanks to everyone. I'll research both the SNDNETF and SAVRST and see which is easier. Thanks for all the info.

Related things commenters have not yet elaborated upon:

Sending things with sndnetf often means you pre-bundle the to-be-sent things into a save file, which you need to restore on the other side. This is two additional (manual) steps per sending.

Sndnetf and friends uses is a store and forward mechanism and was initially meant for distributing small objects such as "office documents", or email. Hence the name SNADST for the base technology.

Worst case you generate a lot of unnecessary I/O by
- Creating a save file on the source
- Calling sndnetf, doing a copy of it
- Calling rcvnetf/wrknetf on the receiver side, which generates a copy of it
- Restoring the save file

This all is mitigated through using savrst* commands ("Object Connect").

Savrst is prone to the same limitations as rst* commands are. You need certain authority. I have found that running savrst* on one machine to another, as qsecofr on the source machine spawns the needed rst* command on the receiving side. For me this worked without the need to enter any passwords or other means to legitimate myself to the other machine. With SNA as transport, on V4R5. I'm not sure which security implications arise from that. Probably one can accidentally or maliciously overwrite stuff on the destination machine, coming from an untrusted source machine tucked secretly under your desk? I have not yet tried this deliberately, and not yet verified if this can only happen if both sides have the same qsecofr password. More insights from others about newer IBM i releases versions might be interesting.

On a side note: Without proper configuration, using SNA as transport, both mechanisms also transfer data over the wire in plain text. As long as the wire is in reality confined to a CEC, I'd not bother.

:wq! PoC