IBM i TR8 virtual Ethernet fix: which path fits your partition

If you installed IBM i 7.5 TR8 or 7.6 TR2 on a partition that reaches the network through virtual Ethernet, IBM now says that install can take the partition completely off the network. The IBM i TR8 virtual Ethernet fix is a pair of HIPER PTFs, MJ11725 for 7.5 and MJ11726 for 7.6. What IBM’s pages do not do is tell you which of the several paths they describe applies to your partition. The most prominent one, removing the TR, costs the most and is the one most shops no longer need.

What IBM flagged, and when

The timeline matters, because the advice changed inside a week.

IBM announced 7.6 TR2 and 7.5 TR8 on July 15, and both became generally available on July 24. Shops have been installing them for two and a half months.

IBM’s first version of the advisory said installing either TR can cause a complete loss of network connectivity on partitions using Virtual Ethernet adapter type 268C. IBM Japan’s translated copy, dated September 28, still preserves what that version told shops, and the guidance was blunt. If you had not installed the TR, consider postponing it. If you had and lost the network, open a case, and the only known recovery was a Licensed Internal Code (LIC) SLIP install back to the pre-TR level. If you were on the TR with no symptoms yet, IBM called it “a business decision based on risk acceptance.”

On October 2, the English advisory (node 7289463) was updated. Development had found the root cause and IBM named two HIPER PTFs: MJ11725 for 7.5 and MJ11726 for 7.6, “expected to be available for download on 10/03/2026.” Both have the TR as a prerequisite. A companion page (node 7289627) laid out three scenarios for removing the TR, with one line above all of them: “There is no need to perform any of the following Scenarios when the fixing PTF is applied to the system.”

Two cautions on reading those pages. The advisory says the PTFs were “released” and “expected” in the same section, so confirm they are orderable with SNDPTFORD or Fix Central before you book the window. And IBM Japan’s page still shows the older, pre-fix guidance.

When TR8 and TR2 shipped, our coverage of the July Technology Refresh said to apply it on your normal patch cycle. For partitions with 268C resources, that advice now needs this HIPER attached to it.

Are you exposed? Check for 268C first

Type 268C is how IBM i reports a virtual Ethernet port. IBM’s documentation says that for each enabled virtual port, the system creates a communications resource “such as CMNxx with a resource type of 268C.” A partition hosted by VIOS owns no physical LAN adapters at all, so every network path it has is a 268C. Internal virtual LANs between partitions on the same frame are 268C too.

That makes the population wide. A small shop with one LPAR that owns its own physical card may be clear. A shop running production, development, and an HA target as VIOS clients is likely exposed on every partition.

Run this on each partition:

SELECT RESOURCE_NAME, DEVICE_TYPE, DEVICE_MODEL, STATUS,
       TEXT_DESCRIPTION, LOCATION_CODE
  FROM QSYS2.HARDWARE_RESOURCE_INFO
 WHERE DEVICE_TYPE = '268C';

Zero rows means the advisory does not apply to that partition. Any rows mean it does. The view arrived in 7.5 TR1, so every system that could have this problem already has it. On a green screen, WRKHDWRSC TYPE(*CMN) shows the same resources with their type.

Do not skip the partitions you rarely think about. The HA target, the dev box, and the partition someone set up for a vendor’s interface all count. If you run IBM i in a hosted or cloud Power environment, run the query rather than assume what the provider did with the network.

Where your PTFs stand

The second query tells you which situation you are in:

SELECT PTF_PRODUCT_ID, PTF_IDENTIFIER, PTF_LOADED_STATUS
  FROM QSYS2.PTF_INFO
 WHERE PTF_IDENTIFIER IN ('MJ99407', 'MJ99408', 'MJ11725',
                          'MJ99501', 'MJ99502', 'MJ11726');

The IDs come from IBM’s two pages. On 7.5, MJ99408 is the PTF that enables TR8, MJ99407 is the TR7 equivalent, and MJ11725 is the fix. On 7.6, MJ99502 enables TR2, MJ99501 is TR1, and MJ11726 is the fix. A status of APPLIED or PERMANENTLY APPLIED on MJ99408 or MJ99502 means the TR is active on that partition. No row for the fix PTF means you do not have it yet.

Which TR8 or TR2 fix path fits your partition

Every 7.5 or 7.6 partition with 268C resources falls into one of these. The right move differs for each, and IBM’s removal page only covers part of the picture.

Your partition	What to do	Expect
A. On 7.5 or 7.6, TR8/TR2 not installed	Install the TR group, current cumulative package, and the fix PTF together in one window. Or hold the TR until you can.	One IPL. You were going to take it for the TR anyway.
B. On TR8/TR2, network working	Apply MJ11725 or MJ11726 at your next window. Do not wait for the next cume.	Plan an IPL window unless the cover letter says it applies immediately.
C. On TR8/TR2, network already gone	Get in through the HMC console, open a case, apply the fix PTF from media or an image the partition can reach without its LAN.	Support involvement. Usually less work than a SLIP install.
D. Upgrading to 7.5 or 7.6 now	Upgrade with current media and cume, and load the fix PTF in the same pass.	Normal upgrade plus one PTF.

Situation B is the one people will put off, because nothing is broken. The earlier version of IBM’s advisory described staying on the TR without the fix as accepting risk, and the impact list on the advisory includes “reduced ability to remotely manage, monitor, or administer the system.” A partition that drops off the network at the next IPL is a bad way to find out you were in B.

Situation C is harder than it looks on paper, because the fix has to reach a partition with no network. If the partition is a VIOS client, a virtual optical device backed by an image on the VIOS is the usual route. Otherwise it is physical media. Open the case first and let IBM support tell you the fastest path for your configuration.

For situation D, IBM’s removal page describes a way to upgrade while skipping the TR: start from an older LIC resave (RSH for TR7 on 7.5, RSC for TR1 on 7.6), then load the current cume while omitting MJ99408 or MJ99502. Unless you have a specific reason to stay off the TR, that path is now more work for less.

Why removing the Technology Refresh is the expensive path

IBM’s removal scenarios were the right answer for the few days when no fix existed. Now they are mostly a trap for people who read the second IBM page before the first.

All the PTFs involved carry the MJ prefix, which IBM uses for Licensed Internal Code. IBM’s PTF FAQ is explicit that LIC PTFs “can only be removed *PERM,” and that a LIC PTF that has been permanently applied “cannot be removed.” Our reading, and it is an inference rather than IBM’s stated reason, is that this is why IBM’s back-out path for the TR is a SLIP install of LIC from SAVSYS media instead of a quick RMVPTF.

That has three costs a shop should see before choosing it:

Scenario 1 needs a SAVSYS taken before you installed the latest PTFs. Plenty of shops apply a cume without a fresh SAVSYS. If you did, that path is closed to you.
Scenarios 2 and 3 need older media you probably cannot download. IBM says earlier resaves, cumes, and group packages are “generally not available through IBM ESS or standard PTF ordering systems.” You open a case to get them.
Omitting the TR-enabling PTF leaves a messy PTF picture. IBM expects some PTFs to fail to load because they require MJ99408 or MJ99502, automatic marking may not happen, and WRKPTFGRP can show your groups as Not Installed afterward. IBM calls that expected.

That last one is the cost nobody prices in. Six months from now, someone looks at WRKPTFGRP, sees Not Installed on groups you deliberately left that way, and spends a day finding out why. If the only person who knows is the one who ran the window, that knowledge leaves when they do. It is the same problem as an undocumented RPG program, and the same fix: write it down while it is fresh. The IBM i AI Field Guide has prompts built for capturing that kind of operational knowledge before it walks out the door.

Removal makes sense in one case: IBM support tells you to, for your configuration. Otherwise apply the fix.

Before you IPL: have a way in that does not use the network

The failure IBM describes removes remote access. So before any window that touches TR8, TR2, or the fix PTFs, confirm you can reach each partition without its LAN.

For most LPARs that means an HMC 5250 console session, tested before the window rather than during the outage. One wrinkle: IBM documents a configuration where the LAN console itself runs over a 268C virtual adapter. If that is how your console is set up, the problem in this advisory could take the console down with everything else. Know which kind of console you have before you start.

Three more things belong in the plan:

Take a SAVSYS before the window. It keeps IBM’s Scenario 1 available as a fallback if the fix PTF misbehaves on your system.
List what depends on each partition’s network. IBM’s impact list includes “service interruptions affecting dependent systems.” An HA target that loses its network stops receiving replication, and ODBC and API clients start failing. Tell their owners the window is happening.
Run the 268C query again after the IPL, and check that the resources report OPERATIONAL. Then test an actual connection, not just a ping.

If you have been burned by a PTF batch before, this is the same pattern as the August security PTFs that broke NetServer mounts and older ACS drivers: IBM ships a change, the change has a side effect, and the warning arrives after shops have already installed it.

Bottom line

Run the 268C query on every IBM i 7.5 and 7.6 partition this week. If it returns nothing, this advisory is not yours. If it returns rows, the answer for almost everyone is the same: get MJ11725 or MJ11726 onto the partition alongside the TR, in one window, with a SAVSYS taken first and a tested HMC console session open.

Do not start down IBM’s TR removal scenarios unless IBM support tells you to. They were the only option for a few days in late September. Since October 2 they are the slow, expensive option, and IBM’s own page says you do not need them once the fix is applied.

Frequently Asked Questions

Does the IBM i TR8 and TR2 virtual Ethernet problem affect physical Ethernet adapters?

IBM’s advisory is scoped to Virtual Ethernet adapter type 268C. A partition that owns its physical adapters and has no virtual Ethernet ports is outside that scope. Many partitions have both, so run the HARDWARE_RESOURCE_INFO query rather than assuming.

Do I need to remove TR8 or TR2 to fix the network loss?

No. IBM’s own removal guidance says none of its removal scenarios are needed once the fixing PTF is applied. Apply MJ11725 on 7.5 or MJ11726 on 7.6, together with the cumulative package and group PTFs that carry the TR.

Does applying MJ11725 or MJ11726 need an IPL?

IBM says systems that are not already on TR8 or TR2 will need an IPL as part of the update, because the TR is a prerequisite. Both are Licensed Internal Code PTFs, so read the cover letter for immediate or delayed apply and book an IPL window either way.

Is IBM i 7.4 affected by the virtual Ethernet advisory?

IBM’s advisory names only IBM i 7.5 TR8 and IBM i 7.6 TR2, and 7.4 is not listed. Watch the numbering: 7.4 has its own TR8, which shipped in 2023, and it is not the release this advisory covers.