cloudstack-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrija Panic <andrija.pa...@gmail.com>
Subject Re: Problems configuring KVM host
Date Wed, 17 Oct 2018 19:42:39 GMT
Huh - so indeed, AFAIK there is no mentioning of the networking bridges in
this quick quide...that's funny.

Please follow the Basic Zone networking guide from this page  (in your case
you don't use vlans as far as I can see):
http://docs.cloudstack.apache.org/en/4.11.1.0/installguide/hypervisor/kvm.html

In short your eth0 should be part of bridge named cloudbr0 (you VMs will be
on your inner network, next to your MGMT server on same subnet, right)?
If so just make "eth0" part of the "cloudbr0" network bridge - and THIS
NAME !!!  (not the IP addresss...) is what is used as KVM TRAFFIC LABEL (in
this case, for both management and guest network) - so once more - name of
the INTERFACE (in this case name of the bridge - that is "cloudbr0") is
what should be set as the KVM traffic label.

The way above will work (just FYI if interesting), CloudStack will search
for physical interface which name matches the "KVM traffic label" and will
join any automatically provisioned VM interfaces (vNIC) to this bridge -
give or take (it works a bit different on Advanced Zone), but I believe
this is true for Basic Zone.

Please give feedback.

Thanks
Andrija

So please fix that - and let me know if that works for you. I will try to
see to update this Quick guide.



On Wed, 17 Oct 2018 at 20:38, Alexandre Bruyere <bruyere.alexandre@gmail.com>
wrote:

> Quick update:
>
> There is something wrong either with the guide, the hardware/network config
> or the community repository - because even after following to the letter
> the guide (which lead the management server to be a host), it still refuses
> to add the host to manage (even though it's localhost)
>
> On Wed, Oct 17, 2018 at 1:46 PM Alexandre Bruyere <
> bruyere.alexandre@gmail.com> wrote:
>
> > To Andrija:
> >
> > I'm not sure what you mean by making the NIC part of the BRIDGE - if you
> > mean make sure it's part of the inner network, it's already done, all
> > machines (physical and virtual) have Internet access.
> > I am looking forward to an updated guide if you do get to it, though (the
> > thought is much appreciated).
> > -----
> > KVM host configurations:
> > Starting point: CentOS 6.10 Minimal Install
> >
> > ifcfg-eth0
> >
> > DEVICE=eth0
> > HWADDR=[Redacted for readability]
> > TYPE=Ethernet
> > UUID=[Redacted for readability]
> > ONBOOT=yes
> > NM_CONTROLLED=no
> > BOOTPROTO=none
> > IPADDR=204.168.1.51
> > NETMASK=255.255.255.0
> > GATEWAY=204.168.1.1
> > DNS1=192.168.101.1
> >
> > SELinux was set to permissive
> >
> > NTP was installed
> >
> > Community package repository was used:
> >
> > cloudstack.repo
> >
> > [cloudstack]
> > name=cloudstack
> > baseurl=http://download.cloudstack.org/centos/6/4.11/
> > enabled=1
> > gpgcheck=0
> >
> >
> > Modified qemu.conf to enable vnc_listen on 0.0.0.0
> >
> > Libvrtd.conf had several lines altered as per the tutorial:
> > listen_tls = 0
> > listen_tcp = 1
> > tcp_port = "16059"
> > auth_tcp = "none"
> > mdns_adv = 0
> >
> > Uncommented LIBVIRTD_ARGS in /etc/sysconfig/libvirtd
> >
> > -----
> > Zone config:
> >
> > Name: CSTestZone
> > Network Type: Basic
> > Public: Yes
> > DNS 1: 192.168.101.1
> > Internal DNS 1: 192.168.101.1
> > Guest Network KVM traffic label: 204.168.1.0/24
> > Management Network KVM traffic label: 204.168.1.0/24
> >
> > -----
> > And for good measure, here's the error that the management server throws
> > when I try adding the host (thanks for the path to the logs!)
> >
> > 2018-10-17 13:41:59,867 WARN  [c.c.h.k.d.LibvirtServerDiscoverer]
> > (qtp1386767190-15:ctx-0e54785c ctx-ee01f7ae) (logid:1a5a07f0)  can't
> setup
> > agent, due to com.cloud.utils.exception.CloudRuntimeException: Failed to
> > setup keystore on the KVM host: 204.168.1.51 - Failed to setup keystore
> on
> > the KVM host: 204.168.1.51
> > 2018-10-17 13:41:59,868 WARN  [c.c.r.ResourceManagerImpl]
> > (qtp1386767190-15:ctx-0e54785c ctx-ee01f7ae) (logid:1a5a07f0) Unable to
> > find the server resources at http://204.168.1.51
> > 2018-10-17 13:41:59,868 INFO  [c.c.u.e.CSExceptionErrorCode]
> > (qtp1386767190-15:ctx-0e54785c ctx-ee01f7ae) (logid:1a5a07f0) Could not
> > find exception: com.cloud.exception.DiscoveryException in error code list
> > for exceptions
> > 2018-10-17 13:41:59,868 WARN  [o.a.c.a.c.a.h.AddHostCmd]
> > (qtp1386767190-15:ctx-0e54785c ctx-ee01f7ae) (logid:1a5a07f0) Exception:
> > com.cloud.exception.DiscoveryException: Unable to add the host
> >         at
> >
> com.cloud.resource.ResourceManagerImpl.discoverHostsFull(ResourceManagerImpl.java:820)
> >         at
> >
> com.cloud.resource.ResourceManagerImpl.discoverHosts(ResourceManagerImpl.java:606)
> >
> >
> > -----------
> > To Simon:
> > Fully plan on migrating to CentOS 7.x or Ubuntu Server LTS - this is
> > merely a test to see if the behavior of Cloudstack is adequate for the
> > project. As Andrija mentioned, the tutorial makes use of CentOS 6.x, and
> > that's what I'm going to until it works, because from experience it's
> > *never* a good idea to start changing things up without understanding how
> > and why software works. Though admittedly that's hypocritical because the
> > guide did make use of the management server as a host as well...
> >
> > On Wed, Oct 17, 2018 at 12:22 PM Andrija Panic <andrija.panic@gmail.com>
> > wrote:
> >
> >> Well, the guide (Quick guide) is targeting CentOS 6.8 AFAIK, I will see
> if
> >> I have some time to update that to CentOS 7 in near future - if that
> makes
> >> sense ?
> >>
> >> On Wed, 17 Oct 2018 at 17:56, Simon Weller <sweller@ena.com.invalid>
> >> wrote:
> >>
> >> > In addition to Andrija's comments below, I'd recommend you use Centos
> >> 7.x
> >> > rather than 6.x, as  Centos/RHEL development is targeted towards that
> >> > release now and has been for a couple of years.
> >> >
> >> >
> >> > - Si
> >> >
> >> > ________________________________
> >> > From: Andrija Panic <andrija.panic@gmail.com>
> >> > Sent: Wednesday, October 17, 2018 10:51 AM
> >> > To: users
> >> > Subject: Re: Problems configuring KVM host
> >> >
> >> > Hi Alexandre,
> >> >
> >> > irrelevant to your explanation above (which is good, to understand
> whole
> >> > setup!), there is something with network configuration on the Zone
> >> level, I
> >> > assume, per the message from the script:
> >> >
> >> >
> >>
> https://github.com/apache/cloudstack/blob/master/agent/bindir/cloud-setup-agent.in#L76
> >> >
> >> >
> >> > Above is an quick-installation link that you followed (all fine! and
> you
> >> > move MGMT to separate server), but that article it has external links
> to
> >> > how to configure KVM host networking - so we don't know how did you
> >> > configure it. In general, the NIC configuration (from the quick guide)
> >> > should be modified so that NIC is part of the BRIDGE (IP parameters
> you
> >> can
> >> > move from NIC to Bridge or not, it should not be relevant for this
> >> sample
> >> > setup) - and make sure that this bridge is defined as the "KVM traffic
> >> > label" while you configured your Zone for probably both the Management
> >> and
> >> > the Guest Network (you can also edit already existing zone - disable
> >> zone,
> >> > edit Guest and Management network (inside physical network, inside
> >> zone) to
> >> > define new KVM traffic label and finally enable zone).
> >> >
> >> > Anyway, please post configuration that you applied on KVM host and
> Zone
> >> > level - since it seems that setup-agent script don't know what bridge
> to
> >> > use for Management and Guest traffic.
> >> > Btw, Management server logs are located
> >> > here: /var/log/cloudstack/management/management-server.log (on the
> >> > management server, obviously :) )
> >> >
> >> > Cheers
> >> >
> >> > On Wed, 17 Oct 2018 at 16:27, Alexandre Bruyere <
> >> > bruyere.alexandre@gmail.com>
> >> > wrote:
> >> >
> >> > > Hello.
> >> > >
> >> > > I'm trying to set up a small Cloudstack setup on a test bench, but
> I'm
> >> > > having issues getting the host/agent running properly.
> >> > >
> >> > > First, here's the configuration of the test bench:
> >> > >
> >> > > Server #1:
> >> > > Hardware - Two active NICs: one that goes out to the company's
> network
> >> > > (we'll call it Outer), the other into a switch on the bench (into
a
> >> > private
> >> > > subnet that we'll call Inner). HP Proliant DG380 G5 (2x quad-core
> >> Intel
> >> > > Xeon - they do not support nested virtualization, 6GB RAM, HP P400
> >> RAID
> >> > > controller running a single SATA HDD)
> >> > > Software: ESXi 5.5. Two virtual routers, one for each NIC. Two
> virtual
> >> > > machines: an Ubuntu VM that acts as a NAT router and bridge between
> >> Inner
> >> > > and Outer (we'll call it RT), and a Cloudstack management server
> >> > > (functional, we'll call it CSMan) connected to Inner.
> >> > >
> >> > > Server #2:
> >> > > Hardware: One active NIC on Inner. Same hardware as Server #1 except
> >> it
> >> > is
> >> > > using a single SAS drive.
> >> > > Software: CentOS 6.10 KVM host configured as per the "KVM Setup and
> >> > > Installation" section of this guide:
> >> > >
> >> > >
> >> >
> >>
> https://cloudstack-documentation.readthedocs.io/en/4.11.1.0/quickinstallationguide/qig.html
> >> > > . NIC is configured with a static Inner IP, gateway is RT, DNS
> servers
> >> > are
> >> > > the company's.
> >> > >
> >> > > Switch: Completely unconfigured Netgear 24-port switch. On the Inner
> >> > > network, there is no DHCP or DNS setup (as the guide I used
> >> specifically
> >> > > says to not add one for the purposes of this setup)
> >> > >
> >> > > Laptop: Simple laptop, Windows 10, configured with an address on
> >> Outer,
> >> > > default gateway is RT. Connects to Server #1 via the vSphere client
> >> using
> >> > > an Outer address, connects to the VMs and Server #2 via PuTTY using
> an
> >> > > Inner address
> >> > >
> >> > >
> >> > > Now onto the problem:
> >> > > CSMan simply fails to add the host, and I can't find any information
> >> on
> >> > why
> >> > > that is (can't find logs). Server #2, on the other hand, fails to
> >> > complete
> >> > > the cloudstack-setup-agent command, returning the error "Failed to
> get
> >> > > default route. Please configure your network ot have a default
> route."
> >> > >
> >> > > Searching the issue only returns a few mail threads that failed to
> >> help,
> >> > > and searching the error only returns installation guides and what
> >> appears
> >> > > to be the source for Cloudstack.
> >> > >
> >> > > Where should I start to even troubleshoot the issue?
> >> > >
> >> > > Thank you very much for your help.
> >> > >
> >> >
> >> >
> >> > --
> >> >
> >> > Andrija Panić
> >> >
> >>
> >>
> >> --
> >>
> >> Andrija Panić
> >>
> >
>


-- 

Andrija Panić

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message