Description of problem: teamd.conf seams be missing in libteam Version-Release number of selected component (if applicable): 1.31 How reproducible: Try to start a network team with nmcli. It gives an error a refuses to bring up team Steps to Reproduce: 1. Install libteam-tools and dependencies (install nmcli); 2. Configure network team with nmcli (e.g nmcli con add type team con-name team0 ; nmcli con add type team-slave ifname $IFNAME master team0 ) 3. Try to start the team (e.g. nmcli connection up team-slave-$IFNAME ; nmcli con up team0 ) 4. You should have some kind of error saying IF team0 is still down. Fixed by copying a teamd.conf file in /etc/dbus-1/system.d/ from el8 teamd-1.31-2 It seams that now nmcli uses DBUS e needs some kind of description. Don't know if already fixed in cauldron. BTW: the man file of teamd.conf does exist in libteam-tools. Thank you in advance. Kernel 6.1.15-desktop GLib 2.66.8
Thank you for the report. Looking on Cauldron (Mageia 9), libteam-tools does include team.conf: $ urpmf teamd.conf libteam-tools:/usr/share/man/man5/teamd.conf.5.xz I will check this on Mageia 8. "Fixed by copying a teamd.conf file in /etc/dbus-1/system.d/ from 'el8 teamd-1.31-2' " Did not understand the last 'from' bit.
CC: (none) => lewyssmith
My previous comment is *wrong*. I overlooked the '/man/man5/' bits! So even Cauldron does not offer 'teamd.conf' itself. Mageia 8: $ urpmf teamd.conf libteam-tools:/usr/share/man/man5/teamd.conf.5.xz Just the man page. But it may be that the file itself should be generated by the commands you cite: nmcli con add type team con-name team0 nmcli con add type team-slave ifname $IFNAME master team0 which would be a NetworkManager problem. Please do explain 'el8 teamd-1.31-2'.
Version: 8 => CauldronSource RPM: libteam-1.31-1.mga8.src.rpm => libteam-1.31-1.mga8.src.rpm, networkmanager-1.26.6-1.1.mga8.src.rpmWhiteboard: (none) => MGA8TOO
Created attachment 13747 [details] teamd.conf file from centOS
Hi. I uploaded the teamd.conf file. I downloaded it from rpm find (CentOS): https://centos.pkgs.org/8/centos-baseos-aarch64/teamd-1.31-2.el8.aarch64.rpm.html I don't think nmcli commands can create it. Sorry I didn't make myself clear. Once I copied teamd.conf in /etc/dbus-1/system.d/ the team started working... Thank you.
Actually I downloaded the x86_64 version.... ;-)
Should be fixed in next libteam-1.31-5.mga9 update!
CC: (none) => geiger.david68210
WOW! Thanks a lot!
Assigning to QA, Packages in 8/Core/Updates_testing: ====================== libteamdctl0-1.31-1.1.mga8 lib64teamdctl0-1.31-1.1.mga8 libteam5-1.31-1.1.mga8 lib64team5-1.31-1.1.mga8 libteam-devel-1.31-1.1.mga8 lib64team-devel-1.31-1.1.mga8 libteam-tools-1.31-1.1.mga8 python3-team-1.31-1.1.mga8 From SRPMS: libteam-1.31-1.1.mga8.src.rpm
Assignee: bugsquad => qa-bugs
Mageia8, x86_64 Taking a look at this. Installed the core release files and copied the attached config file to /etc/dbus-1/system.d/ $ nmcli connection up team-slave-eno1 ; nmcli con up team0 Error: unknown connection 'team-slave-eno1'. Connection successfully activated (master waiting for slaves) (D-Bus active path: /org/freedesktop/NetworkManager/ActiveConnection/27) It is not going to work even after updating because it does not recognise my network adapter. I tried setting the environment variable IFNAME and used that but that did not work either. ???
CC: (none) => tarazed25
Removed the teamd.conf file and updated to lib64teamdctl0-1.31-1.1.mga8 lib64team5-1.31-1.1.mga8 lib64team-devel-1.31-1.1.mga8 libteam-tools-1.31-1.1.mga8 python3-team-1.31-1.1.mga8 $ cd /etc/dbus-1/system.d/ $ ls -l teamd.conf -rw-r--r-- 1 root root 409 Jan 9 2020 teamd.conf $ nmcli con add type team con-name team0$ nmcli con add type team con-name team0 Warning: There are 2 other connections with the name 'team0'. Reference the connection by its uuid 'e8d90b3e-af0f-4b1a-8f9e-dad2278c512f' Connection 'team0' (e8d90b3e-af0f-4b1a-8f9e-dad2278c512f) successfully added. $ nmcli con add type team-slave ifname $IFNAME master team0 $ nmcli con up team0 Connection successfully activated (master waiting for slaves) (D-Bus active path: /org/freedesktop/NetworkManager/ActiveConnection/31) At least the config file is being supplied now. Looks like the decks needed to be cleared before starting this test. This looks OK as far as it goes.
Whiteboard: MGA8TOO => MGA8TOO MGA8-64-OK
Note that the `nmcli con add type team con-name team0` command was run once. The doubling in comment 10 is a typo. And maybe team1 would have been a better idea.
(In reply to Len Lawrence from comment #9) > Mageia8, x86_64 > Taking a look at this. > > Installed the core release files and copied the attached config file to > /etc/dbus-1/system.d/ > > $ nmcli connection up team-slave-eno1 ; nmcli con up team0 > Error: unknown connection 'team-slave-eno1'. > Connection successfully activated (master waiting for slaves) (D-Bus active > path: /org/freedesktop/NetworkManager/ActiveConnection/27) > > It is not going to work even after updating because it does not recognise my > network adapter. I tried setting the environment variable IFNAME and used > that but that did not work either. ??? My two bits.... Given eth0 and eth1 the names of the "real" eth cards and "team0" the name of the team, the following commands worked for me: [root@ ~]# nmcli con show (just to be sure of the names....) [root@ ~]# nmcli con down eth0 [root@ ~]# nmcli con down eth1 [root@ ~]# nmcli con add type team con-name team0 [ config '{"device": "team0", "runner": {"name": "loadbalance"}, "ports": {"team-slave-eth0": {}, "team-slave-eth1": {}}}' ] (the part after "team0" is optional) [root@ ~]# nmcli con add type team-slave ifname team-slave-eth0 master team0 [root@ ~]# nmcli con add type team-slave ifname team-slave-eth1 master team0 (this part is also optional. I modified the ifcfg file by hand) [root@ ~]# nmcli con mod team0 ipv4.method manual [root@ ~]# nmcli con mod team0 ipv4.addresses xxx.xxx.xxx.xxx/24 [root@ ~]# nmcli con mod team0 ipv4.gateway xxx.xxx.xxx.xxx (to bring the team up, you should "up" the slaves first) [root@ ~]# nmcli con up team-slave-eth0 [root@ ~]# nmcli con up team-slave-eth1 [root@ ~]# nmcli con up team0 Thank you again!
(In reply to David GEIGER from comment #6) > Should be fixed in next libteam-1.31-5.mga9 update! Does that mean that this should now be relabeled as a Mageia 8 bug?
CC: (none) => andrewsfarm
Now that Cauldron has been fixed, yes. I have just done that. It was raised for M8, but since I found that it applied also to M9, I changed it to Cauldron + M8. Now it is M8 only. Just to confirm M9 is fixed (compare comment 2): $ urpmf teamd.conf libteam-tools:/etc/dbus-1/system.d/teamd.conf *** libteam-tools:/usr/share/man/man5/teamd.conf.5.xz Thanks to DavidG for doing this.
Whiteboard: MGA8TOO MGA8-64-OK => MGA8-64-OKVersion: Cauldron => 8CC: lewyssmith => (none)
I thought that was probably the case, but it's better to ask and be sure. Validating.
Keywords: (none) => validated_updateCC: (none) => sysadmin-bugs
CC: (none) => davidwhodginsKeywords: (none) => advisory
An update for this issue has been pushed to the Mageia Updates repository. https://advisories.mageia.org/MGAA-2023-0031.html
Status: NEW => RESOLVEDResolution: (none) => FIXED