Todo #13236
closedDocument link speed limitations with igc and ix on 6100/4100
100%
Description
The I225 built-in phy doesn't support fixed operation, so a speed/duplex setting is emulated by selecting that single value in the autonegotiation menu (this is what the Linux driver does).
NG 6069#note-5
This effectively means the igc ports will not work with explicitly set speeds at the other end. The behavior is observed with ix ports on the 6100 as well.
Updated by Marcos M almost 3 years ago
- Subject changed from Document link speed limitations with igc and ix to Document link speed limitations with igc and ix on 6100/4100
Updated by Jim Pingle almost 3 years ago
Has that been tested and confirmed? I remember some discussion around that back when the driver was first brought in, but I don't recall if anyone confirmed it to be the case and I'm not turning up anything specific when searching on slack.
Updated by Marcos M almost 3 years ago
It's been tested by a customer, along with SW and CL.
At best, the ix and igc ports on both the 6100 and 4100 will fail back to autoselect if they are hard-set, linking at whichever speed is auto-negotiated. If the other side is hard-set, the ports will fail in different ways, but the result is the same - traffic will not flow.
Updated by Chris Linstruth almost 3 years ago
Which makes this language on the pfSense interface config pages, though correct, probably in need of some adjustment for these devices:
Explicitly set speed and duplex mode for this interface. WARNING: MUST be set to autoselect (automatically negotiate speed) unless the port this interface connects to has its speed and duplex forced.
Updated by Jim Pingle almost 3 years ago
- Status changed from In Progress to Feedback
- % Done changed from 0 to 100
Note added: https://gitlab.netgate.com/docs/pfsense-platforms/-/commit/05b5a002a9ecf6915681b90c4f75bf49248dcc1d
https://docs.netgate.com/pfsense/en/latest/solutions/netgate-4100/io-ports.html#networking-ports
https://docs.netgate.com/pfsense/en/latest/solutions/netgate-6100/io-ports.html#networking-ports
Updated by Marcos M almost 3 years ago
Note looks good, however when we tested this, the issue happened on the ix
ports as well (see #note-3).
Updated by Jim Pingle almost 3 years ago
Marcos M wrote in #note-8:
Note looks good, however when we tested this, the issue happened on the
ix
ports as well (see #note-3).
All ix or just ix on 4100/6100? If only on 4100/6100, is it just the copper/RJ45 combo ports, or all ix ports on the 6100? Since the SFP would be locked to the speed of the module it's not clear how it might behave if the fiber module supports multiple speeds on the 10G SFP ports of the 6100. I'm curious if anyone has testing the behavior on SFP with a 1G/10G module on both ends and trying to force it to 1G. The GUI does offer a choice of limiting the media speed like it does for traditional ports.
Updated by Marcos M almost 3 years ago
For ix, it's the same on the 4100/5100/6100. I only see testing history on the RJ-45 ix ports, not SFP.
Updated by Jim Pingle almost 3 years ago
This should be better now:
https://gitlab.netgate.com/docs/pfsense-platforms/-/commit/ce3288bce4b70211f1296deddaa2c51a76c3927b
While there I also added notes about copper SFPs:
https://gitlab.netgate.com/docs/pfsense-platforms/-/commit/2e583df39e401b2821d109b5359c47deee46c248
(Which replaced some items in an alternate branch I cleaned up).