Re: [xcat-user] unexpected hostname

Bin XA Xu Wed, 13 Mar 2019 19:49:03 -0700

Hi Sandra,

By default, xcat will not set the static hostname on compute node, but leave it updated by dhcp client on compute nodes.

And xcat will generate the dhcp lease for the node (for the provision NIC ), so that compute node could get its new name from dhcp server when it boot again.

```

    host mid21tor24cn06 {
dynamic;
hardware ethernet <MAC>;
uid <MAC>;
fixed-address 172.16.24.6;
        supersede server.ddns-hostname = "mid21tor24cn06";
        supersede host-name = "mid21tor24cn06";
        supersede conf-file =
                "http://172.16.253.100:80/tftpboot/petitboot/mid21tor24cn06";
}

```

Normally, there should be only 1 dhcp nic enabled on compute nodes, so the `hostnamectl` would be shown as

```

hostnamectl
Static hostname: localhost.localdomain
Transient hostname: mid21tor24cn06.mydomain.com

```

But from your node, it seems the public NIC still get another hostname (does that name `dev2-pub` exists in xCAT MN's dhcp lease file?)

Another option to xCAT is to set static IP on the provision NIC, it means to add `confignics -s` or `confignetwork -s`

Then it will try to set the static hostname into `/etc/hostname`. or `/etc/sysconfig/network`, then every time you change the hostname, you need to update compute node accordingly.

And the output will be like below:

```

# hostnamectl
   Static hostname: mid21tor24cn06
         Icon name: computer
        Machine ID: ...
           Boot ID: ...
Operating System: Red Hat Enterprise Linux Server 7.6 (Maipo)
       CPE OS Name: cpe:/o:redhat:enterprise_linux:7.6:GA:server
            Kernel: Linux 4.14.0-115.el7a.ppc64le
      Architecture: ppc64-le
# cat /etc/hostname
mid21tor24cn06

```

But it might still be updated the `Transient name` via dhcp client, so I'm not if it works for you.

Bin Xu

HPC Software Development
Software Defined Infrastructure, IBM Systems

Phone: 86-010-82454067

E-mail: [email protected]

----- Original message -----
From: Carl <[email protected]>
To: xCAT Users Mailing list <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Thu, Mar 14, 2019 6:09 AM

Hi Sandra,

I cant speak to the xCat way of setting the hostname, as it just works for us.

Its possible that NetworkManager is setting your hostname based on DHCP, see this https://access.redhat.com/solutions/2816951 for more information.

Cheers,

Carl.

On Thu, 14 Mar 2019 at 08:39, Sandra Maksimovic <[email protected]> wrote:

Hi Bin,

Please note that I am currently testing without any postscript modification to the NICs since I would prefer to get it working this way. Right now it’s just building with the information provided by xCAT.

Here is the requested output:

# hostnamectl

   Static hostname: localhost.localdomain

Transient hostname: dev2-pub.meerkat.mcri.edu.au

         Icon name: computer-server

           Chassis: server

        Machine ID: 721ed6c3cf8144149d7671556f14c705

           Boot ID: e3031b42f7234a5792510c5d61656869

Operating System: CentOS Linux 7 (Core)

       CPE OS Name: cpe:/o:centos:centos:7

            Kernel: Linux 3.10.0-862.14.4.el7.x86_64

      Architecture: x86-64

Do you know how a node would select its hostname normally?

Thanks,

Sandra

From: Bin XA Xu <[email protected]>
Sent: Wednesday, 13 March 2019 4:34 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [xcat-user] unexpected hostname

Hi Sandra,

     You have mentioned that there is a postscripts to modify eth3 network config file to enable DHCP get the its IP from your organization dhcp server, right?

     Does this organization dhcp server offer a hostname for the IP?

      BTW: what's output of `hostnamectl` on node `dev2`?

Bin Xu

HPC Software Development
Software Defined Infrastructure, IBM Systems

Phone: 86-010-82454067

E-mail: [email protected]

----- Original message -----
From: Sandra Maksimovic <[email protected]>
To: "[email protected]" <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Wed, Mar 13, 2019 1:06 PM

Hi there,

Unfortunately this issue has reared its ugly head again – this time on a physical host.

I’ve noticed a pattern, basically the host will always apply the DNS name used for our ‘pub’ public network when I want it to default to the vanilla ‘dev2’ mgmt net name. Here is an example of entries from the DNS:

dev2        A     10.40.114.200     (eth0 – configured as dhcp on physical host, handed out by xCAT’s DHCP server)

dev2-pub    A     172.16.14.200     (eth3 – deployed as dhcp on physical host, handed out by our organisation’s DHCP server)

dev2-data   A     10.50.114.200     (eth4 – deployed as static on physical host)

dev2-obm    A     10.60.114.200     (eth5 – deployed as static on physical host)

For some reason the server is always being deployed as ‘dev2-pub’. I know I can get around this by using a postbootscript to overwrite the xCAT applied config, but I’d prefer to not do this because I’d rather not have to manage so many scripts.

Is there some way I can ‘force’ xCAT to prefer a particular hostname over another for a host?

Please let me know if you need any more info!

Thanks,

Sandra

From: Bin XA Xu <[email protected]>
Sent: Friday, 2 November 2018 5:37 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [xcat-user] unexpected hostname

Sandra,

   Thanks for the information, we saw that you are using the esx to mange your VMs. We don't have such environment, and Yuan tried to recreate your problem on our own KVM based environment, but the issue is not seen.

   Whatever I'm glad that you've made a workaround successfully, and Thanks for sharing your experience here.

Regards,

Bin Xu

HPC Software Development
Software Defined Infrastructure, IBM Systems

Phone: 86-010-82454067

E-mail: [email protected]

----- Original message -----
From: Sandra Maksimovic <[email protected]>
To: "[email protected]" <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Fri, Nov 2, 2018 11:37 AM

Hi Bin,

In this case I’m only providing the MAC for eth0 for the mgt0 node, not for any others. This is a special case to get around the fact that I’m deploying to a VM which by nature does not have OBM capabilities, but since I have access to the hypervisor directly this is a non-issue.

On xCAT I’ve also only set one specific NIC for DHCP (site.dhcpinterfaces=ens224) so it definitely should not be trying to hand out IPs to any other NICs other than eth0 on the mgt0 node (both eth0 and ens224 are configured to use the 10.40.0.0 management/provisioning network).

Many thanks,

Sandra

From: Bin XA Xu <[email protected]>
Sent: Friday, 2 November 2018 1:32 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [xcat-user] unexpected hostname

Sandra,

    So you have a public DHCP server which listening on eth1's subnet, which can assign the right public IP and hostname (mgt0, instead of mgt0-pub) .

    It is okay to workaround this.

    Normally, for xCAT deployed nodes, only provision nic will be set to DHCP, other's are configured as static. And the provision NIC will get the hostname(mgt0) from DHCP server on xCAT MN/SN. And Grep it just like below:

```

     grep boston01 /var/lib/dhcpd/dhcpd.leases
host boston01 {
        supersede server.ddns-hostname = "boston01";
        supersede host-name = "boston01";
        supersede conf-file = "http://10.6.29.1/tftpboot/petitboot/boston01";

```

    I'm wondering that your dhcp serer on xCAT MN/SN also has the lease for mgt0's eth1 if you also provide its MAC in mac table.

    If this the case, it is better to disable xCAT's dhcp serer listening on eth1 as you have already an Corporate external DHCP server.



You can set the site.dhcpinterfaces to which nics you want to listening and `makedhcp -n` to regenerate the xcatd dhcp configuration, `man site` for more detail information on that.

Bin Xu

HPC Software Development
Software Defined Infrastructure, IBM Systems

Phone: 86-010-82454067

E-mail: [email protected]

----- Original message -----
From: Sandra Maksimovic <[email protected]>
To: "[email protected]" <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Fri, Nov 2, 2018 6:59 AM

Hi Yuan,

I ended up with the following config (there’s probably a bit of redundancy here):

Object name: mgt0

    arch=x86_64

    authdomain=mcri.edu.au

    chain=standby

    conserver=xcat

    currchain=boot

    currstate=boot

    domaintype=activedirectory

    groups=mgt,vm

    ip=10.40.113.99

    mac=<snip>

    mgt=esx

    netboot=pxe

    nfsdir=/install

    nfsserver=xcat

    nicaliases.eth0=mgt0

    nicaliases.eth1=mgt0-pub

    nicaliases.eth2=mgt0-data

    nichostnamesuffixes.eth0=-mgmt

    nichostnamesuffixes.eth1=-pub

    nichostnamesuffixes.eth2=-data

    nicips.eth0=10.40.113.99

    nicips.eth1=172.16.13.99

    nicips.eth2=10.50.113.99

    nicnetworks.eth0=Management

   nicnetworks.eth1=Service

    nicnetworks.eth2=Data

    nictypes.eth0=Ethernet

    nictypes.eth1=Ethernet

    nictypes.eth2=Ethernet

    os=centos7.5

    otherinterfaces=-pub:172.16.13.99,-data:10.50.113.99

    ou=<snip>

    postbootscripts=otherpkgs

    postscripts=syslog,remoteshell,syncfiles,setupntp,confignics,MCRI_MgtSetup

    profile="">

    provmethod=centos7-mgt

    routenames=14NetRoute,MySQLUCSCRoute

    servicenode=xcat

    status=booted

    statustime=11-01-2018 15:07:59

    updatestatus=synced

    updatestatustime=11-01-2018 14:06:16

I’m also now running a postscript to set eth1 to receive its reserved IP from our public DHCP server as a work around and this seems to do what I’d like in terms of setting the hostname correctly.

Cheers,

Sandra

From: Yuan Y Bai <[email protected]>
Sent: Thursday, 1 November 2018 6:49 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [xcat-user] unexpected hostname

Hi Sandra,

I define mgt0 node in my MN, delete `mgt0-pub:172.16.13.99... ...` in `otherinterfaces` from hosts table, and add `nicaliases.eth0=mgt0-mgt` in node definition, then execute `makehosts mgt0`, it can generate `mgt0-pub` and `mgt0-data` in /etc/hosts file, the domain `cluster.com` is coming from my `site` table. I curious that you need to define mgt0-pub and mgt0-data in otherinterfaces from `hosts` table.

BTW: is there DHCP server for eth1 before the node provision?

My example here:

[root@bybc0607 ~]# lsdef mgt0
Object name: mgt0
    arch=x86_64
    authdomain=mcri.edu.au
    chain=standby
    conserver=xcat
    currchain=boot
    currstate=boot
    domaintype=activedirectory
    groups=mgt,vm
    hostnames=mgt0
    ip=10.40.113.99
    mac=<snip>
    mgt=esx
    netboot=pxe
    nfsdir=/install
    nfsserver=xcat
    nicaliases.eth0=mgt0-mgt
    nichostnamesuffixes.eth2=-data
    nichostnamesuffixes.eth1=-pub
    nicips.eth2=10.50.113.99
    nicips.eth1=172.16.13.99
    nicips.eth0=10.40.113.99
    nicnetworks.eth2=Data
    nicnetworks.eth1=Public
    nicnetworks.eth0=Management
    nictypes.eth2=Ethernet
    nictypes.eth1=Ethernet
    nictypes.eth0=Ethernet
    os=centos7.5
    ou=<snip>
    postbootscripts=otherpkgs,<snip>
    postscripts=syslog,remoteshell,syncfiles,setupntp,confignics,<snip>
    profile="">    provmethod=centos7-mgt
    routenames=14NetRoute,MySQLUCSCRoute
    servicenode=xcat
    status=failed
    statustime=10-29-2018 14:14:01
    updatestatus=failed
    updatestatustime=10-29-2018 13:53:40
[root@bybc0607 ~]# makehosts mgt0

[root@bybc0607 ~]# grep mgt0 /etc/hosts
10.40.113.99 mgt0 mgt0.cluster.com mgt0-mgt
10.50.113.99 mgt0-data mgt0-data.cluster.com
172.16.13.99 mgt0-pub mgt0-pub.cluster.com

[root@bybc0607 ~]# lsxcatd -v
Version 2.14.4 (git commit 51bd7fea2746d1812aa0eba3d655d63e16b718e2, built Wed Oct 17 06:15:55 EDT 2018)

Best Regards
--------------------------------------------------
Yuan Bai (白媛)

CSTL HPC System Management Development
Tel:86-10-82451401
E-mail: [email protected]
Address: IBM ZGC Campus. Ring Building 28,
ZhongGuanCun Software Park,No.8 Dong Bei Wang West Road, Haidian District,
Beijing P.R.China 100193

IBM环宇大厦
北京市海淀区东北旺西路8号，中关村软件园28号楼
邮编:100193

----- Original message -----
From: Sandra Maksimovic <[email protected]>
To: 'xCAT Users Mailing list' <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Thu, Nov 1, 2018 1:29 PM

Btw I managed to work around this issue by setting eth1 to use DHCP and eth0 to send DHCP_HOSTNAME using a postscript.

Cheers,

Sandra

From: Sandra Maksimovic
Sent: Tuesday, 30 October 2018 6:30 PM
To: 'xCAT Users Mailing list' <[email protected]>
Subject: RE: [xcat-user] unexpected hostname

Hi Yuan,

Just to let you know, it seems that when I remove otherinterfaces=”mgt0-pub:172.16.13.99,mgt0-data:10.50.113.99” from the mgt0 definition, the /etc/hosts file does not regenerate with mgt0-pub or mgt0-data entries, only mgt0 and its fqdn is listed.

The xcat servicenode should be managing nodes over the 10.40.0.0/24 network, however, I don’t think this has been setup properly because the servicenode table is blank. A lot of this new cluster’s configuration has been carried over from our current prod iteration so I’m not sure whether some of these definitions are still relevant.

The /var/lib/dhclient directory is missing the dhclient.leases file but contains the following:

# cat chrony.servers.eth0

10.40.115.100 iburst

# cat ntp.conf.predhclient.eth0

<blank>

The IP 10.40.115.100 is the management NIC on my xCAT server, which seems to indicate the correct provisioning network…

I’ve just noticed that when I run ‘dhclient’ manually on the ‘mgt0-pub’ node the leases file appears along with some others…

dhcp-server-identifier on eth0 (which is the mgt/provisioning NIC on the 10.40.0.0 net) is 10.40.115.100

host-name is “mgt0”

I’m now wondering what would have stopped this information from being generated during deployment? And would this have managed to impact the hostname?

Many thanks,

Sandra

From: Yuan Y Bai <[email protected]>
Sent: Monday, 29 October 2018 4:40 PM
To: [email protected]
Cc: [email protected]
Subject: Re: [xcat-user] unexpected hostname

Hi Sandra

From your node definition, `nichostnamesuffixes.eth1=-pub nicips.eth1=172.16.13.99` will generate `172.16.13.99 mgt0-pub ......` entry in /etc/hosts file. No need to `mgt0-pub:172.16.13.99` in otherinterfaces.

And you use service node, `servicenode=xcat`, which network service node use?

Could you login `mgt0-pub` and check lease file under directory `/var/lib/dhclient` to see what are `dhcp-server-identifier` and `host-name`?

It seems `mgt0` node get hostname `mgt0-pub` from 172.xx.xx.xx DHCP server. The provision network should 10.xx.xx.xx network.

Best Regards
--------------------------------------------------
Yuan Bai (白媛)

CSTL HPC System Management Development
Tel:86-10-82451401
E-mail: [email protected]
Address: IBM ZGC Campus. Ring Building 28,
ZhongGuanCun Software Park,No.8 Dong Bei Wang West Road, Haidian District,
Beijing P.R.China 100193

IBM环宇大厦
北京市海淀区东北旺西路8号，中关村软件园28号楼
邮编:100193

----- Original message -----
From: Sandra Maksimovic <[email protected]>
To: "'[email protected]'" <[email protected]>
Cc:
Subject: Re: [xcat-user] unexpected hostname
Date: Mon, Oct 29, 2018 12:15 PM

Hi Bin,

Thanks for your response.

mgt0 and mgt0-pub do not point to the same IP address nor are they in the same subnet. Please see the output below:

Object name: mgt0

    arch=x86_64

    authdomain=mcri.edu.au

    chain=standby

    conserver=xcat

    currchain=boot

   currstate=boot

    domaintype=activedirectory

    groups=mgt,vm

    hostnames=mgt0

    ip=10.40.113.99

    mac=<snip>

    mgt=esx

    netboot=pxe

    nfsdir=/install

    nfsserver=xcat

    nichostnamesuffixes.eth0=-mgmt

    nichostnamesuffixes.eth1=-pub

    nichostnamesuffixes.eth2=-data

    nicips.eth0=10.40.113.99

    nicips.eth1=172.16.13.99

    nicips.eth2=10.50.113.99

    nicnetworks.eth0=Management

    nicnetworks.eth1=Public

    nicnetworks.eth2=Data

    nictypes.eth0=Ethernet

    nictypes.eth1=Ethernet

    nictypes.eth2=Ethernet

    os=centos7.5

    otherinterfaces=mgt0-pub:172.16.13.99,mgt0-data:10.50.113.99

    ou=<snip>

    postbootscripts=otherpkgs,<snip>

    postscripts=syslog,remoteshell,syncfiles,setupntp,confignics,<snip>

    profile="">

    provmethod=centos7-mgt

    routenames=14NetRoute,MySQLUCSCRoute

    servicenode=xcat

    status=failed

    statustime=10-29-2018 14:14:01

    updatestatus=failed

    updatestatustime=10-29-2018 13:53:40

FYI some of our postscripts are failing during deployment which is why the updatestatus=failed.

Also, thanks Brian for your suggestion, I shall look into this further regarding the NIC setup. I did a quick test and this doesn’t appear to be what I’m after at this stage since the deployed node’s hostname is unaffected when specifying the nicaliases.

Thanks,

Sandra

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐

On Friday, October 26, 2018 5:17 PM, Bin XA Xu <[email protected]> wrote:

Hi Sandra,

    Is the mgt0 and mgt0-pub pointing to the same IP address, or in the same subnet? And what's your `mgt01` definition, you can use `lsdef mgt01` to get the information and hide the sensitive attributes.

    And Yuan, do you have more suggestions?

Bin Xu

HPC Software Development
Software Defined Infrastructure, IBM Systems

Phone: 86-010-82454067

E-mail: [email protected]

----- Original message -----

From: Sandra Maksimovic via xCAT-user <[email protected]>

To: "[email protected]" <[email protected]>

Cc: Sandra Maksimovic <[email protected]>

Subject: [xcat-user] unexpected hostname

Date: Thu, Oct 25, 2018 11:35 PM

Hi all,

xCAT/HPC/list newbie here!

I have recently configured an xCAT node and am attempting to provision a separate management node, but for some reason xCAT is sort of not applying the expected hostname.

I'd like the resulting hostname on the node to just be "mgt0", but instead it's tacking on the public NIC suffix as well as the FQDN, i.e. mgt0-pub.meerkat.mcri.edu.au

The cluster is entirely CentOS7 based and will be eventually utilising MOAB and PBS/Torque for scheduling and resource management. The version of xCAT for this particular build is v2.14.4.

I've trawled through the debug enabled build logs and stepped through post.rh.common and from what I can tell the node should just be named "mgt0" (sans all suffixes).

Also, the DNS on the xCAT node contains entries for "mgt0", "mgt0-data", "mgt0-pub", but (if this is indeed the issue) I'm not sure why xCAT would have selected "mgt0-pub" to hand out when the node is being provisioned via its management IP which is actually associated with "mgt0" (as opposed to its public one which is associated with "mgt0-pub").

Any ideas on other avenues that might be worth investigating?

Also, please feel free recommend some useful resources for learning xCAT and/or HPC in general! I'm already heavily utilising the official xCAT docs and the Sourceforge Wiki/mailing list search...

Cheers,

Sandra

Sent from ProtonMail, encrypted email based in Switzerland.

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

This e-mail and any attachments to it (the "Communication") are, unless otherwise stated, confidential, may contain copyright material and is for the use only of the intended recipient. If you receive the Communication in error, please notify the sender immediately by return e-mail, delete the Communication and the return e-mail, and do not read, copy, retransmit or otherwise deal with it. Any views expressed in the Communication are those of the individual sender only, unless expressly stated to be those of Murdoch Children’s Research Institute (MCRI) ABN 21 006 566 972 or any of its related entities. MCRI does not accept liability in connection with the integrity of or errors in the Communication, computer virus, data corruption, interference or delay arising from or in respect of the Communication.

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

_______________________________________________
xCAT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/xcat-user

Re: [xcat-user] unexpected hostname

Reply via email to