Here is an example of a host autocreated by Fact import on SLES 11: Suse SUSE Linux Enterprise Server 11 (x86_64). Since architecture is part of the operating system, and every operating system can have multiples, it is very confusing to show this. There are two issues here: First, that it doesn't show the minor release number (aka service pack), which is significant. Second is that it shows the architecture, while that major/minor combo can actually have multiple architectures (e.g. x86_64, i386, even s390x).

The vital parts of the OS structure (osfamily, major, minor) are all set correctly. It is only the description that is confusing.

I would suggest instead that it say something more like "SLES <major version> SP <minor version>" for the description.

We can (and have) worked around this issue by changing the descriptions.

We discussed this in IRC yesterday:

4:12:28 PM) mmoll: mjolnir40k: tell us more :)
(4:12:52 PM) jsherrill is now known as jsherrill_afk
(4:15:09 PM) mjolnir40k: We have a mixture of RHEL and SLES boxes, running on x86_64 and s390x architectures
(4:15:40 PM) mjolnir40k: On SLES 11, the x86_64 is being assigned to the Operating System tagged s390x with architecture
(4:16:05 PM) mjolnir40k: in seems to be fact related, since we manually re-assigned it and when puppet runs it goes back to the s390x operating system
(4:19:17 PM) hjensas [~hjensas@c80-216-37-143.bredband.comhem.se] entered the room.
(4:19:17 PM) mmoll: mjolnir40k: if you run "facter -p" on such a box the output looks OK? if tick both architectures in the OS is it better then?
(4:19:18 PM) ayoung is now known as ayoung_dad_mode
(4:19:54 PM) ***mmoll just saw today hat both, SLES11 SP2 and SP3 have "SUSE Linux Enterprise Server 11 (x86_64)" as description...
(4:20:07 PM) mjolnir40k: Yes, the boxes report the right architectures for facter
(4:20:41 PM) mjolnir40k: Our installation has been through Foreman 1.2, 1.3 and 1.4
(4:20:51 PM) mmoll: mjolnir40k: do you have something in the audit logs when such a change occurs?
(4:21:30 PM) mjolnir40k: Getting one now.
(4:22:26 PM) jlabocki left the room (quit: Ping timeout: 240 seconds).
(4:22:43 PM) tmclaugh[work] left the room (quit: Quit: tmclaugh[work]).
(4:22:46 PM) jlabocki [~Adium@174.47.44.30] entered the room.
(4:23:11 PM) simon-AS5591 left the room (quit: Quit: Leaving.).
(4:23:39 PM) mmoll: mjolnir40k: what is changed? the description of the OS or the architecture of the host?
(4:25:31 PM) mjolnir40k: The description of the os has "s390x" in it, not x86_64 as we expected
(4:25:34 PM) YamakasY_ left the room (quit: Remote host closed the connection).
(4:25:48 PM) mjolnir40k: We also have an OS with x86_64 in it, but it is not being assigned to that operating system
(4:25:50 PM) mmoll: mjolnir40k: OK, that's what I thought...
(4:26:15 PM) mmoll: mjolnir40k: open an issue for that, please, I'll look into that. I think 1.4 introduced that behavour
(4:26:18 PM) Sonixblaze left the room (quit: Quit: Leaving).
(4:27:13 PM) taloszerg left the room (quit: Remote host closed the connection).
(4:27:14 PM) mmoll: mjolnir40k: it's taking lsbdistdescription from facter. The problem is that this string in SUSE is also containing the architecture
(4:27:58 PM) mmoll: and not the version, which is leading to my problem with SP2 vs. SP3 using the same string
(4:29:09 PM) mjolnir40k: OK. I will open an issue on this. Can you give a quick pointer to the code where this is done? I'm new to reading Rails code (and ruby in general) and having a hard time tracking it down
(4:29:51 PM) colinmcnamara left the room (quit: Quit: My MacBook Pro has gone to sleep. ZZZzzz…).
(4:30:27 PM) mmoll: mjolnir40k: app/services/facts_parser.rb probably
(4:31:55 PM) mjolnir40k: Thanks
(4:32:20 PM) mjolnir40k: Ah, I see a bunch of stuff changed in there for 1.4
(4:32:46 PM) mjolnir40k: I have another foreman installation with Debian boxes and the handling for it changed too - I see where and why now