UEFI Boot Mode

[Dietmar Rieder]

Hi all,

first of all thanks for all your work on Rocks, it is very much appreciated!

I just tried to install Rocks 7 Beta on a HP DL580 Gen9 Server via the iLO 4 virtual drive (as usb).

It seems that the iso can not be booted in UEFI Boot mode as it is the case with a standard CentOS 7 iso.

When I switch to one time boot in "Legacy Boot Mode" I can install Rocks 7, however it won't boot after installation.

Are there any plans to get this running with UEFI, just as the standard CentOS distribution does?

Thanks

  Dietmar

[Philip Papadopoulos]

surprising.. I'll have to test this. It's a uefi hybrid iso.  At least it is supposed to be. I'm using Lorax, which is the tool that Centos uses to create boot media.

-P

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.
To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/16ff111d-f7d0-41d8-b4ae-aa87be5239fe%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Philip Papadopoulos, Ph.D

ppapad...@ucsd.edu

(858) 822-3628

[Dietmar Rieder]

I have to add that when I set the boot mode after installing rocks again to legacy, the installed rocks boots from the HDD.

Dietmar

[Philip Papadopoulos]

Dietmar,

I have a fix (I need to figure out the root cause -- something I'm doing wrong)

Please go to beta7.rocksclusters.org and download the rocks-installer-7-0.iso image.

This will boot in a UEFI-only BIOS system.

BUT,  you will need to add some information once the installer comes up

AFTER you have configured your network, you need to go to the Software Sources Screen.

you will need to enter a network-based repo:

Please choose:

http://mirror.centos.org/centos/7/os/x86_64

After that, you will need to visit the software selection screen.  There is no choice to be made there,

but it must be visited.

The rocks-installer-7.0.iso image is what comes out of the Lorax process.   Based on that image, a next step creates
the kernel roll iso.. The kernel ISO has a local repo defined, which anaconda picks up automatically.  However, I'm not creating

the UEFI version of that correctly (That's what I need to debug).

NOTE: don't forget the /boot partition when booting in UEFI mode. :-)

Thanks for the report,

Phil

On Tue, Sep 12, 2017 at 1:40 PM, Dietmar Rieder <dietmar...@gmail.com> wrote:

I have to add that when I set the boot mode after installing rocks again to legacy, the installed rocks boots from the HDD.

Dietmar

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/2819f47e-6bed-44bc-9585-498d143d56f1%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Dietmar Rieder]

Hi Philip,

the fix works, I can now boot and install and reboot the head node in UEFI boot mode, however the compute-nodes can not be booted and installed over PXE in UEFI boot mode. I have to set them to one time Legacy boot mode, then the installer is launched. The installer stops at the "Installation Summary" screen and wants me to do a software selection and I have to manually select the "Minimal Install". Furthermore I have to specify the installation source. It is set to http://127.0.0.1/install/rocks/x86_64 so I have to change 127.0.0.1 to the IP address of the head node (10.0.0.100 in my case). Partitioning is also not done "automagically".

Dietmar

[Philip Papadopoulos]

The second parts sounds very wrong, can you see if both the tracker and httpd are running on your frontend. (rocks-tracker). 

P

--

You received this message because you are subscribed to the Google Groups "rocks7-beta" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/7331be07-a193-4e41-93ba-4f7472dd0a59%40googlegroups.com.

[Philip Papadopoulos]

Now that I'm not on my phone.

The installer needs to retrieve the kickstart file, and it should point to http://localhost/fetchRocksKS.py.   That is a cgi running locally on the installing node that will

ask the frontend (repeatedly, if needed) for a kickstart file. 

IF that stage fails, it will return a very basic kickstart file so that contact with the node isn't lost. 

If your webserver on the frontend is not on, you will end up in the state that your system is in.

Once, the kickstart file is retrieved, files are downloaded via the rocks-tracker. This is for scalable installation.  

If you want to bypass both of these

set the installaction ofa node to "install notracker"

and set UseTracker to False

# rocks set host installaction <nodename> action="install notracker"

# rocks set host attr <nodename> UseTracker False

-P

--

You received this message because you are subscribed to the Google Groups "rocks7-beta" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/7331be07-a193-4e41-93ba-4f7472dd0a59%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Dietmar Rieder]

....as far as I see...

# ps ax | grep htt
2651 ?        Ss     0:00 /usr/sbin/httpd -DFOREGROUND
3186 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
3189 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
3192 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
3194 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
3197 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
20421 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND
59217 ?        S      0:00 /usr/sbin/httpd -DFOREGROUND

# ps ax | grep track
6157 ?        S      0:00 /opt/rocks/bin/tracker-server

....they seem to be running

Dietmar

Am Mittwoch, 13. September 2017 16:51:02 UTC+2 schrieb Philip Papadopoulos:

The second parts sounds very wrong, can you see if both the tracker and httpd are running on your frontend. (rocks-tracker). 

P

On Sep 13, 2017 7:45 AM, "Dietmar Rieder" <dietmar...@gmail.com> wrote:

Hi Philip,

the fix works, I can now boot and install and reboot the head node in UEFI boot mode, however the compute-nodes can not be booted and installed over PXE in UEFI boot mode. I have to set them to one time Legacy boot mode, then the installer is launched. The installer stops at the "Installation Summary" screen and wants me to do a software selection and I have to manually select the "Minimal Install". Furthermore I have to specify the installation source. It is set to http://127.0.0.1/install/rocks/x86_64 so I have to change 127.0.0.1 to the IP address of the head node (10.0.0.100 in my case). Partitioning is also not done "automagically".

Dietmar

Am Dienstag, 12. September 2017 18:57:27 UTC+2 schrieb Dietmar Rieder:

Hi all,

first of all thanks for all your work on Rocks, it is very much appreciated!

I just tried to install Rocks 7 Beta on a HP DL580 Gen9 Server via the iLO 4 virtual drive (as usb).

It seems that the iso can not be booted in UEFI Boot mode as it is the case with a standard CentOS 7 iso.

When I switch to one time boot in "Legacy Boot Mode" I can install Rocks 7, however it won't boot after installation.

Are there any plans to get this running with UEFI, just as the standard CentOS distribution does?

Thanks

  Dietmar

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta...@googlegroups.com.

[Philip Papadopoulos]

Ok. On the installing node check the contents of /run/install/ks.cfg and let me know what it looks like

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/0985e729-0cab-4b40-8a12-f4b0e84c5547%40googlegroups.com.

[Dietmar Rieder]

See the attached screenshot.

Moreover, in the http ssl_accessl_og on the head node I see:

10.0.0.254 - - [13/Sep/2017:17:39:49 +0200] "POST /install/ens7f0 HTTP/1.1" 404 212
10.0.0.254 - - [13/Sep/2017:17:40:03 +0200] "POST /install/ens7f0 HTTP/1.1" 404 212                     
10.0.0.254 - - [13/Sep/2017:17:40:21 +0200] "POST /install/ens7f0 HTTP/1.1" 404 212                     
10.0.0.254 - - [13/Sep/2017:17:40:34 +0200] "POST /install/ens7f0 HTTP/1.1" 404 212                     
10.0.0.254 - - [13/Sep/2017:17:40:45 +0200] "GET /install/ens7f0.FAILED HTTP/1.1" 404 219

(ens7f0 is the private network device on the head node, strange that the compute node is asking for a file named ens7f0)

Dietmar

[Philip Papadopoulos]

Output of 

rocks list host profile <nodename>

There is an error generating the Kickstart file. 

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/5d351162-c07f-42d3-a747-d639f3c232cb%40googlegroups.com.

[Dietmar Rieder]

Output attached...

Dietam

[Cooper, Trevor]

Dietmar,

How did you add your nodes to the Rocks DB on the frontend?

Did you use insert-ethers or the contents of previous rocks dump?

Your installing compute node is reporting back six (6) ethernet devices detected...

eno49 - 3c:a8:2a:ed:72:e8
eno50 - 3c:a8:2a:ed:72:e9
eno51 - 3c:a8:2a:ed:72:ea
eno52 - 3c:a8:2a:ed:72:eb
eno57 - 48:df:37:03:13:8c
eno58 - 48:df:37:03:13:8d

... and, oddly and as you've noted, none match the what is in the POST/GET from /var/log/httpd/ssh_access_log which should look like...

10.1.1.251 - - [10/Sep/2017:22:02:38 -0700] "POST /install/sbin/kickstart.cgi HTTP/1.1" 200 132900
10.1.1.251 - - [10/Sep/2017:22:28:18 -0700] "GET /install/sbin/public/setDbPartitions.cgi HTTP/1.1" 200 1

If you used the contents of previous Rocks dump it may be better to add a single node with insert-ethers, let it install and see how CentOS detects and Rocks records the interfaces in the DB then replicate the interface naming for the rest of your similar nodes.

I found that attempting to specify (ie. override) the default interface naming doesn't ever work as expected (from CentOS 6) and it was quicker to simply accept the new naming convention.

Also, it seems like some of your Rocks DB attributes may be messed up. Can you provide (sanitize if you wish/need) the output of...

# rocks list attr | egrep "Kickstart|Info" | sort

The POST above for the kickstart is built from some of these attributes, namely...

Kickstart_PrivateKickstartBasedir: install
Kickstart_PrivateKickstartCGI: sbin/kickstart.cgi

If these are somehow borked that will explain the inability to get the kickstart file for the node. HOW they got borked is what needs to be determined.

Worst case, if you used output from a previous (ie. Rocks 6.x) rocks dump to 'restore' the configuration in your Rocks 7 cluster (which is essentially what a restore roll will do) you may have to reinstall your frontend. 'Restore' isn't yet working...

Trevor

> To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/c7176afb-15ae-49d1-bb52-3b27c564cf4d%40googlegroups.com.

> For more options, visit https://groups.google.com/d/optout.

> <host_profile_compute-0-0.txt>

[Philip Papadopoulos]

This commit:

https://github.com/rocksclusters/core/commit/38778684ce180639e966b98857e797e554d18e9f

Should fix the UEFI build bug so that kernel rolls are also UEFI-bootable.

> To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

> To post to this group, send email to rocks...@googlegroups.com.
> Visit this group at https://groups.google.com/group/rocks7-beta.
> To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/c7176afb-15ae-49d1-bb52-3b27c564cf4d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

> <host_profile_compute-0-0.txt>

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/0390156B-BD49-486E-91F5-D5D0226A4D65%40sdsc.edu.

For more options, visit https://groups.google.com/d/optout.

--

[Philip Papadopoulos]

As trevor points out, you aren't getting a kickstart request properly, which might indicate that you have changed the bootaction.

If you look in /tftpboot/pxelinux/pxelinux.cfg, you should see files that are the HEX address of the IP addresses of nodes.

Can you find the one that is for this particular node, and send the contents

It will look something like this  (note, the install action for this particular node (compute-0-0, ip address 10.1.116.254)  is "install notracker"

[root@mordor ~]# cd /tftpboot/pxelinux/pxelinux.cfg/

[root@mordor pxelinux.cfg]# ls

0A0174FC  0A0174FD  0A0174FE  default

[root@mordor pxelinux.cfg]# cat 0A0174FE

default rocks

prompt 0

label rocks

        localboot 0

[root@mordor pxelinux.cfg]# rocks set host boot action=install compute-0-0

[root@mordor pxelinux.cfg]# cat 0A0174FE

default rocks

prompt 0

label rocks

        kernel vmlinuz-7.0-x86_64

        append inst.ks.sendmac inst.sshd lang= devfs=nomount selinux=0 ipv6.disable=1 rocks=client ksdevice=bootif inst.repo=http://10.1.116.1/install/rocks-dist/x86_64 inst.updates=http://10.1.116.1/install/rocks-dist/x86_64/images/updates.img inst.ks=https://10.1.116.1/install/sbin/kickstart.cgi inst.noverifyssl ip=10.1.116.254 gateway=10.1.116.1 netmask=255.255.255.0 dns=10.1.116.1 nextserver=10.1.116.1 initrd=initrd.img-7.0-x86_64

        ipappend 2

-P

On Wed, Sep 13, 2017 at 9:55 AM, Cooper, Trevor <tco...@sdsc.edu> wrote:

> To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

> To post to this group, send email to rocks...@googlegroups.com.
> Visit this group at https://groups.google.com/group/rocks7-beta.
> To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/c7176afb-15ae-49d1-bb52-3b27c564cf4d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

> <host_profile_compute-0-0.txt>

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/0390156B-BD49-486E-91F5-D5D0226A4D65%40sdsc.edu.

For more options, visit https://groups.google.com/d/optout.

--

[Dietmar Rieder]

Hi,

the entire setup (HW and SW) is new, so I did not use any contents of previous rocks installations. And yes, I used insert-ethers to add the nodes.

There are 6 ethernet devices on the nodes (1x 4-port  1Gb and 1x 2-port 10Gb).

MAC addresses and devices seem to be correct:

    eno49 - 3c:a8:2a:ed:72:e8 
    eno50 - 3c:a8:2a:ed:72:e9 
    eno51 - 3c:a8:2a:ed:72:ea 
    eno52 - 3c:a8:2a:ed:72:eb 
    eno57 - 48:df:37:03:13:8c 
    eno58 - 48:df:37:03:13:8d 

Here is the output of:

#  rocks list attr | egrep "Kickstart|Info" | sort  

Info_CertificateCountry:           xx                           
Info_CertificateLocality:          xxx                    
Info_CertificateOrganization:      xxx                         
Info_CertificateState:             xx                        
Info_ClusterContact:               dietmar...@xxxx.xxx   
Info_ClusterLatLong:               N xx.xx, W xx.xx     
Info_ClusterName:                  cluster-02                   
Info_ClusterURL:                   http://xxx.at               
Info_GoogleOTPRoot:                yes                          
Info_GoogleOTPUsers:               yes                          
Kickstart_DistroDir:               /export/rocks                
Kickstart_Keyboard:                us                           
Kickstart_Lang:                    en_US.UTF-8                  
Kickstart_Langsupport:             en_US.UTF-8                  
Kickstart_PrivateAddress:          10.0.0.100                   
Kickstart_PrivateDNSDomain:        local                        
Kickstart_PrivateDNSServers:       10.0.0.100                   
Kickstart_PrivateGateway:          10.0.0.100                   
Kickstart_PrivateHostname:         zeus                         
Kickstart_PrivateInterface:        ens7f0                       
Kickstart_PrivateKickstartBasedir: install                      
Kickstart_PrivateKickstartCGI:     ens7f0                       
Kickstart_PrivateKickstartHost:    10.0.0.100                   
Kickstart_PrivateMTU:              1500                         
Kickstart_PrivateNetmask:          255.255.255.0                
Kickstart_PrivateNetmaskCIDR:      24                           
Kickstart_PrivateNetwork:          10.0.0.0                     
Kickstart_PrivateNTPHost:          zeus.local                   
Kickstart_PrivateSyslogHost:       10.0.0.100                   
Kickstart_PublicAddress:           172.16.xxx.100               
Kickstart_PublicDNSDomain:         icbi.local                   
Kickstart_PublicDNSServers:        172.16.yyy.yyy,192.168.yyy.yyy 
Kickstart_PublicFQDN:              zeus.icbi.local              
Kickstart_PublicGateway:           172.16.xxx.xxx               
Kickstart_PublicHostname:          zeus                         
Kickstart_PublicInterface:         ens7f1                       
Kickstart_PublicKickstartHost:     central.rocksclusters.org    
Kickstart_PublicMTU:               1500                         
Kickstart_PublicNetmask:           255.255.255.0                
Kickstart_PublicNetwork:           172.16.xxx.0                 
Kickstart_PublicNTPHost:           pool.ntp.org                 
Kickstart_Timezone:                Europe/Berlin

It seems that "Kickstart_PrivateKickstartCGI: ens7f0" is not what you expect. I guess this should be fixed, I don't know how this got set?!

Thanks

  Dietmar

[Dietmar Rieder]

Hi,

I get something similar but different:

# rocks set host boot action=install compute-0-0

# cat 0A0000FE

default rocks
prompt 0
label rocks
       kernel vmlinuz-7.0-x86_64

       append inst.ks.sendmac inst.sshd lang= devfs=nomount selinux=0 ipv6.disable=1 rocks=client ksdevice=bootif inst.repo=http://127.0.0.1/install/rocks-dist/x86_64 inst.ks=http://localhost/fetchRocksKS.py inst.updates=http://127.0.0.1/install/rocks-dist/x86_64/images/updates.img rocks.ks=https://10.0.0.100/install/ens7f0 tracker.trackers=10.0.0.100 ip=10.0.0.254 gatew
ay=10.0.0.100 netmask=255.255.255.0 dns=10.0.0.100 nextserver=10.0.0.100 initrd=initrd.img-7.0-x86_64
       ipappend 2

Note the 127.0.0.1 which is different from your output.

Dietmar

> To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/c7176afb-15ae-49d1-bb52-3b27c564cf4d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

> <host_profile_compute-0-0.txt>

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta...@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/0390156B-BD49-486E-91F5-D5D0226A4D65%40sdsc.edu.

For more options, visit https://groups.google.com/d/optout.

[Dietmar Rieder]

Hi,

I just noted that the 127.0.0.1 thing is related to the tracker as Philip mentioned.

Anyway, I now changed the bootaction "install" to the following:

rocks add bootaction action=install kernel=vmlinuz-7.0-x86_64 ramdisk=initrd.img-7.0-x86_64 args="inst.ks.sendmac inst.sshd lang= devfs=nomount selinux=0 ipv6.di

sable=1 rocks=client ksdevice=bootif inst.repo=http://127.0.0.1/install/rocks-dist/x86_64 inst.ks=http://localhost/fetchRocksKS.py inst.updates=http://127.0.0.1/install/rocks-dist/x86_64/

images/inst.updates=http://127.0.0.1/install/rocks-dist/x86_64/images/updates.img rocks.ks=https://10.0.0.100/install/sbin/kickstart.cgi tracker.trackers=10.0.0.100"


... replacing "rocks.ks=https://10.0.0.100/install/ens7f0" with "rocks.ks=https://10.0.0.100/install/sbin/kickstart.cgi"

Now the node seems install correctly (however not being able to boot in UEFI mode after installation).

I have no idea why Kickstart_PrivateKickstartCGI is set to "ens7f0". I can not remember that during the head node setup/configuration I had to set something related to this.

When there is an installation iso with the UEFI boot problem fixed, I'll try to do a completely fresh installation of the head and compute nodes.

Dietmar

[Dietmar Rieder]

Sounds good, thanks, whenever there is a new iso ready, I'll try it.

Dietmar

> To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/c7176afb-15ae-49d1-bb52-3b27c564cf4d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

> <host_profile_compute-0-0.txt>

--
You received this message because you are subscribed to the Google Groups "rocks7-beta" group.

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta...@googlegroups.com.
To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/0390156B-BD49-486E-91F5-D5D0226A4D65%40sdsc.edu.

For more options, visit https://groups.google.com/d/optout.

[Dietmar Rieder]

I just noted that CentOS 7.4 was released today:

https://lists.centos.org/pipermail/centos-announce/2017-September/022532.html

I guess you'll upgrade rocks to it before releasing a new iso....

Dietmar

[Philip Papadopoulos]

Can you send me the anaconda.log file from your frontend install.  Should be in / var/log/anaconda

Also can you look up the correct kickstart directive to install a uefi bootloader... I don't have pure uefi cluster to test.

P

On Sep 13, 2017 11:55 PM, "Dietmar Rieder" <dietmar...@gmail.com> wrote:

Hi,

the entire setup (HW and SW) is new, so I did not use any contents of previous rocks installations. And yes, I used insert-ethers to add the nodes.

There are 6 ethernet devices on the nodes (1x 4-port  1Gb and 1x 2-port 10Gb).

MAC addresses and devices seem to be correct:

    eno49 - 3c:a8:2a:ed:72:e8 
    eno50 - 3c:a8:2a:ed:72:e9 
    eno51 - 3c:a8:2a:ed:72:ea 
    eno52 - 3c:a8:2a:ed:72:eb 
    eno57 - 48:df:37:03:13:8c 
    eno58 - 48:df:37:03:13:8d 

Here is the output of:

#  rocks list attr | egrep "Kickstart|Info" | sort  
Info_CertificateCountry:           xx                           
Info_CertificateLocality:          xxx                    
Info_CertificateOrganization:      xxx                         
Info_CertificateState:             xx                        

Info_ClusterContact:               dietmar.rieder@xxxx.xxx   

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/e66c545d-e9f3-4ae7-a95f-a4de0407e56e%40googlegroups.com.

[Dietmar Rieder]

sure, I attached the log (I just changed IP addresses, e-mail and locations)

Dietmar

[Dietmar Rieder]

I'm not 100% sure what you mean with " look up the correct kickstart directive to install a uefi bootloader"

I hope this is what you are looking for:

I did a standard CentOS 7 installation on a UEFI system and checked the contents of anaconda-ks.cfg


I think this are the relevant parts for installing the bootloader

[...]

bootloader --append=" crashkernel=auto" --location=mbr --boot-drive=sda

# Partition clearing information

clearpart --drives=sda --initlabel

# Disk partitioning information

part /boot/efi --fstype="efi" --ondisk=sda --size=200 --fsoptions="umask=0077,shortname=winnt"

part / --fstype="xfs" --ondisk=sda --size=50000

[... more partitions here ...]

HTH

  Dietmar

[Philip Papadopoulos]

I think it is the /boot/efi partition. I'll look at this when back from vacation. But you could try a custom partitioning to see if that fixes it. 

It might also be the bootloader line.

Which is also customizable

To unsubscribe from this group and stop receiving emails from it, send an email to rocks7-beta+unsubscribe@googlegroups.com.

To post to this group, send email to rocks...@googlegroups.com.
Visit this group at https://groups.google.com/group/rocks7-beta.

To view this discussion on the web visit https://groups.google.com/d/msgid/rocks7-beta/322615c2-c3c2-4149-bac6-9c1dd4d603cc%40googlegroups.com.

[Philip Papadopoulos]

I think this -might- be working now.

beta6.rocksclusters.org has updated kernel, core, base rolls.

There were a couple of items:

1. autopartitioning had to add /boot/efi as a biosboot partition

2. had to teach rocks how to pxe boot a uefi node.

      --- for uefi nodes, the loader needs to be shim.efi

                              , the configuration needs to be a grub2 configuration

      --- for bios nodes, the loader needs to be pxelinux.0

                               , the configuration needs to be syslinux compatible

How this is supposed to work,

all pxe configuration is in /tftpboot/pxelinux

the subdirectory pxelinux.cfg has the correct "stuff" for bios-booting nodes

the subdirectory uefi.cfg has the correct "stuff" for uefi-booting nodes.

When rocks set host boot action=[os|install] is executed both uefi and bios compatible boot files are written. This is because we don't

know what the bios will be.

Please let me know what you find. I have only one system that is pure uefi (pxe-capable as uefi). So my testing is very very limited.

Thanks,

Phil

[Dietmar Rieder]

I'll test this on Monday and report back!

Thanks
Dietmar