On 10/03/2014 12:38 PM, Darr247 wrote:

On 03 October 2014 @13:53 zulu, Digimer wrote:

...

On 03/10/14 09:12 AM, Richer, Mark (CIV) wrote:

...

All,

I am trying to understand better how you give an interface a more descriptive name and get it all working without a reboot, if possible.

I actually wrote a small tutorial on how to do just this.

https://alteeve.ca/w/Changing_Ethernet_Device_Names_in_EL7_and_Fedora_15%2B

I think you missed the "without a reboot" part. :)

Supposedly you can restart udev and then networkservices

Of course, I found a page with a solution that worked right after sending the last email. However, without pointing to me to search for pages in google on reloading udev rules, I wouldn’t have found this so thanks to the list for the collaborative solution.

http://askubuntu.com/questions/82470/what-is-the-correct-way-to-restart-udev

My comments are in [ ]s under some of the steps the author wrote.

This excerpt had the magic sauce I needed — it added a step to modprobe the driver module for the 10g NIC before reloading the udev rules.

The author wrote:

You have to combine all the advice given here in the right order:

1. Bring down the network service networking stop [Comment: I skipped this, on CentOS 7, I would use systemctl stop network but it stopped already and I couldn’t restart it — however, the interfaces worked, maybe because NetworkManager is also still running?] 2. Unload the driver module from the kernel * Find the name of the module lspci -v and look for "Kernel driver in use:" * modprobe -r <driver module> [This was critical. I did lspci -v to find the 10G card and saw the driver module is called ixgbe]. 3. Reload the udev rules udevadm control --reload-rules 4. Trigger the new rules udevadm trigger 5. Load driver modprobe <driver module> 6. Restart the network service networking start

[Very interesting but now when I did systemctl restart network, it worked. The modprobe commands, at least the first one, was critical to both allowing the service to start up again and to change the device name I see in ifconfig and elsewhere.]

1. (optional) Re-run any iptables scripts that referenced the eth interface name before it was up.

I suspect either step 4 or step 5 isn't really needed, but these steps worked for me. You could check after step 4 with step 2.1 to see if the trigger command already did step 5, edit this answer to reflect your findings if you do.

[Above is the poster’s comments … maybe I’ll try to skip 4 and 5 on the next server I execute this procedure on, but I need to leave so I’ll let you know later.]

Cheers! Mark

MARK H RICHER, MS CS NPS-NCR Digital Forensics Lab IT Manager Computer Science Department Naval Postgraduate School - National Capital Region (NCR) 900 N Glebe Rd, Rm 5-182, Arlington, VA 22203 571.858.3254 (o) 571.303.9498 (m) mhricher@nps.edumailto:mhricher@nps.edu

On Oct 3, 2014, at 3:11 PM, Richer, Mark (CIV) <mhricher@nps.edumailto:mhricher@nps.edu> wrote:

Thanks to everyone who responded. This led to some interesting reading and learning, but it hasn’t avoided the reboot.

I found this page on udev: How to reload udev rules without reboot?http://unix.stackexchange.com/questions/39370/how-to-reload-udev-rules-without-reboot http://unix.stackexchange.com/questions/39370/how-to-reload-udev-rules-witho...

Sounds perfect for my question, but at least one server I tried all the suggestions on, it didn’t change anything. A reboot is a “magic sauce,” but it’s nice to know how to avoid this with servers. If I find another solution that works for me, I’ll post it.

Mark

MARK H RICHER, MS CS NPS-NCR Digital Forensics Lab IT Manager Computer Science Department Naval Postgraduate School - National Capital Region (NCR) 900 N Glebe Rd, Rm 5-182, Arlington, VA 22203 571.858.3254 (o) 571.303.9498 (m) mhricher@nps.edumailto:mhricher@nps.edu

On Oct 3, 2014, at 1:16 PM, Robert Moskowitz <rgm@htt-consult.commailto:rgm@htt-consult.com> wrote:

On 10/03/2014 12:38 PM, Darr247 wrote: On 03 October 2014 @13:53 zulu, Digimer wrote: On 03/10/14 09:12 AM, Richer, Mark (CIV) wrote: All,

I am trying to understand better how you give an interface a more descriptive name and get it all working without a reboot, if possible.

I actually wrote a small tutorial on how to do just this.

https://alteeve.ca/w/Changing_Ethernet_Device_Names_in_EL7_and_Fedora_15%2B

I think you missed the "without a reboot" part. :)

Supposedly you can restart udev and then networkservices

_______________________________________________ CentOS mailing list CentOS@centos.orgmailto:CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

On Fri, Oct 3, 2014 at 11:08 PM, Robert Moskowitz rgm@htt-consult.com wrote:

On 10/03/2014 03:11 PM, Richer, Mark (CIV) wrote:

...

Thanks to everyone who responded. This led to some interesting reading and learning, but it hasn’t avoided the reboot.

I found this page on udev: How to reload udev rules without reboot?<http://unix. stackexchange.com/questions/39370/how-to-reload-udev-rules-without-reboot

...

http://unix.stackexchange.com/questions/39370/how-to-reload- udev-rules-without-reboot

Sounds perfect for my question, but at least one server I tried all the suggestions on, it didn’t change anything. A reboot is a “magic sauce,” but it’s nice to know how to avoid this with servers. If I find another solution that works for me, I’ll post it.

I learned this to control the MAC address so that the IPv6 suffix for my servers was more to my liking and I could use an RA prefix.

Mark

...

MARK H RICHER, MS CS NPS-NCR Digital Forensics Lab IT Manager Computer Science Department Naval Postgraduate School - National Capital Region (NCR) 900 N Glebe Rd, Rm 5-182, Arlington, VA 22203 571.858.3254 (o) 571.303.9498 (m) mhricher@nps.edumailto:mhrich er@nps.edu

On Oct 3, 2014, at 1:16 PM, Robert Moskowitz <rgm@htt-consult.commailto: rgm@htt-consult.com> wrote:

On 10/03/2014 12:38 PM, Darr247 wrote: On 03 October 2014 @13:53 zulu, Digimer wrote: On 03/10/14 09:12 AM, Richer, Mark (CIV) wrote: All,

I am trying to understand better how you give an interface a more descriptive name and get it all working without a reboot, if possible.

I actually wrote a small tutorial on how to do just this.

https://alteeve.ca/w/Changing_Ethernet_Device_Names_in_EL7_ and_Fedora_15%2B

I think you missed the "without a reboot" part. :)

Supposedly you can restart udev and then networkservices

---

CentOS mailing list CentOS@centos.orgmailto:CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

Of course if you don't mind rebooting the system, this will work as well: http://wiki.centos.org/FAQ/CentOS7#head-31ebc6642958a0df12304d6aab9a49034a3b...

[root@entos7 ~]# ifconfig

eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

inet 10.30.1.100 netmask 255.255.255.0 broadcast 10.30.1.255

inet6 fe80::a00:27ff:fe54:1d2d prefixlen 64 scopeid 0x20<link>

ether 08:00:27:54:1d:2d txqueuelen 1000 (Ethernet)

RX packets 64 bytes 7690 (7.5 KiB)

RX errors 0 dropped 0 overruns 0 frame 0

TX packets 74 bytes 11580 (11.3 KiB)

TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

inet 172.16.154.10 netmask 255.255.255.0 broadcast 10.30.1.255

inet6 fe80::a00:27ff:fe54:1d3d prefixlen 64 scopeid 0x20<link>

ether 08:00:27:54:1d:3d txqueuelen 1000 (Ethernet)

RX packets 64 bytes 7690 (7.5 KiB)

RX errors 0 dropped 0 overruns 0 frame 0

TX packets 74 bytes 11580 (11.3 KiB)

TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

I really don't get it. Why get into so much fuss just to rename your interfaces????

On Sun, Oct 5, 2014 at 5:18 PM, Eliezer Croitoru eliezer@ngtech.co.il wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

Hey Mark,

You can use "ip" tools to do the trick. For Ubuntu I wrote this upstart script that helps with it without touching udev. You can see it here: http://www1.ngtech.co.il/paste/1175/

You can run this function at runtime and it will change the interface name.

On CentOS you will need to find the right "spot" in boot to apply this link. Before NetworkManager and after udev.

All The Bests, Eliezer

On 10/03/2014 04:12 PM, Richer, Mark (CIV) wrote:

...

All,

I am trying to understand better how you give an interface a more descriptive name and get it all working without a reboot, if possible.

We have 1G and 10G interfaces, and I’m trying to use names like 1G-internal, 1G-external, 10G-private, etc. When I boot up, it’s all fine, but if I add one I’m not sure if there is a way to avoid the reboot? For example, I added the 10G interface names this week.

Specifically, is there a way to change the network interface name you see in ifconfig and nmcli connection without rebooting CentOS 7?

I changed the name in network-scripts. I tried to restart NetworkManager.

I brought down the interface and tried to rename the file and bring it up again, but it still retains the previous run-time setting associated with the same UUID in the file.

Also I find that on all but one of the server on which I did this, I can restart NetworkManager, but network.service is failing to restart. Do I want both active? And if yes, is this indicative of a problem related to changing the interfaces that goes away (only) by rebooting? For some reason, after doing this on several systems, on only one I can restart network.service, but it also still shows the old interface name.

thanks, Mark

MARK H RICHER, MS CS NPS-NCR Digital Forensics Lab IT Manager Computer Science Department Naval Postgraduate School - National Capital Region (NCR) 900 N Glebe Rd, Rm 5-182, Arlington, VA 22203 571.858.3254 (o) 571.303.9498 (m) mhricher@nps.edumailto:mhricher@nps.edu

_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1

iQEcBAEBAgAGBQJUMVMdAAoJENxnfXtQ8ZQUqL8H+wU+CSfoH6gAjJQUN2QaDO8t Rc+/zY2aAvP7vJ1rZtiEdt4he+wRWYCsV7olb2wq2/G+i+8FT14huRpYVufyMwLR VpN/sknu1KNc5fuIY6MTGYUoGDgcgCe3zy77e4cqo5IvPUNyPNfh2+X5H5xwfYaq pvQi6za9U2Gnlj3D65pCwI5svBWbPVNkdQO83xXNlyBhKqEqV1utqG4G5uiOusoM 6EIZZ9lNlLO/3PqUkg6d77HTowsT73qOmAa4KT46iZ1mnB/pibwglt8rxpAieDOD 3G3QScjuOje3EXdnBE6oJ90h6I645X12sAdEOSTITsgdfjAEZiNTIPil6ZaEzrw= =YSN7 -----END PGP SIGNATURE----- _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

On Sun, Oct 5, 2014 at 11:20 PM, Digimer lists@alteeve.ca wrote:

To answer this question using my use-case;

I build HA clusters, and I want to make sure that physical port X on all nodes have the same device name. Biosdevname tries to address this, but doesn't work all the time.

Further, in my case, I've got a minimum of six interfaces in each node, paired into three bonded groups. Having the device name reflect the purpose in the node is very helpful 12~24 months down the road when I need to fix a network problem.

There *are* cogent arguments for renaming interfaces.

cheers

No argument here. I just though that it is much easier to simply pass the following options to grub --> "net.ifnames=0 biosdevname=0" Then you manually edit the interface(s) name.