Change network of ovirt-nodes of a self-hosted engine cluster
Good day! I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine 3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms Does it make sense? -- С уважением, Костырев Александр, системный администратор
Il 25/03/2015 14:32, Kostyrev Aleksandr ha scritto:
Good day!
I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine
I think you should use Hosted Engine global maintenance while doing this. If for some reason the gateway configured in hosted_engine.conf become unreachable I'm not sure it won't try to migrate the VM. Adding Martin to the loop.
3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms
Does it make sense?
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
Il 30/03/2015 08:40, Sandro Bonazzola ha scritto:
Il 25/03/2015 14:32, Kostyrev Aleksandr ha scritto:
Good day!
I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine
I think you should use Hosted Engine global maintenance while doing this. If for some reason the gateway configured in hosted_engine.conf become unreachable I'm not sure it won't try to migrate the VM. Adding Martin to the loop.
Also if for some reason the storage become unavailable during the network settings update, the engine VM will be paused due to missing storage. I'm not sure about what will happen with the SSL certificates, didi?
3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms
Does it make sense?
If you manage to end successfully the migration it would be really nice to have a wiki page describing the steps you took for this task.
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
----- Original Message -----
From: "Sandro Bonazzola" <sbonazzo@redhat.com> To: "Kostyrev Aleksandr" <kostyrev@tutu.ru>, "Users" <users@ovirt.org>, "Martin Sivak" <msivak@redhat.com>, "Yedidyah Bar David" <didi@redhat.com>, "Fabian Deutsch" <fabiand@redhat.com> Sent: Monday, March 30, 2015 9:44:05 AM Subject: Re: [ovirt-users] Change network of ovirt-nodes of a self-hosted engine cluster
Il 30/03/2015 08:40, Sandro Bonazzola ha scritto:
Il 25/03/2015 14:32, Kostyrev Aleksandr ha scritto:
Good day!
I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine
I think you should use Hosted Engine global maintenance while doing this. If for some reason the gateway configured in hosted_engine.conf become unreachable I'm not sure it won't try to migrate the VM. Adding Martin to the loop.
Also if for some reason the storage become unavailable during the network settings update, the engine VM will be paused due to missing storage. I'm not sure about what will happen with the SSL certificates, didi?
In 3.5+, the certs all use the hostnames, which I understand are not intended to be changed, so no issue there. If some of the hosts were deployed with <= 3.4 and upgraded, the certs will use the IP address, and this will not change unless you manually reinstall (re-deploy?). Didn't try that. I guess at least some things will fail.
3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms
Does it make sense?
If you manage to end successfully the migration it would be really nice to have a wiki page describing the steps you took for this task.
Indeed! Best, -- Didi
Hi, the steps make good sense indeed, but use global maintenance mode as Sandro suggested. Good luck and keep us posted. -- Martin Sivák msivak@redhat.com Red Hat Czech RHEV-M SLA / Brno, CZ ----- Original Message -----
----- Original Message -----
From: "Sandro Bonazzola" <sbonazzo@redhat.com> To: "Kostyrev Aleksandr" <kostyrev@tutu.ru>, "Users" <users@ovirt.org>, "Martin Sivak" <msivak@redhat.com>, "Yedidyah Bar David" <didi@redhat.com>, "Fabian Deutsch" <fabiand@redhat.com> Sent: Monday, March 30, 2015 9:44:05 AM Subject: Re: [ovirt-users] Change network of ovirt-nodes of a self-hosted engine cluster
Il 30/03/2015 08:40, Sandro Bonazzola ha scritto:
Il 25/03/2015 14:32, Kostyrev Aleksandr ha scritto:
Good day!
I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine
I think you should use Hosted Engine global maintenance while doing this. If for some reason the gateway configured in hosted_engine.conf become unreachable I'm not sure it won't try to migrate the VM. Adding Martin to the loop.
Also if for some reason the storage become unavailable during the network settings update, the engine VM will be paused due to missing storage. I'm not sure about what will happen with the SSL certificates, didi?
In 3.5+, the certs all use the hostnames, which I understand are not intended to be changed, so no issue there.
If some of the hosts were deployed with <= 3.4 and upgraded, the certs will use the IP address, and this will not change unless you manually reinstall (re-deploy?). Didn't try that. I guess at least some things will fail.
3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms
Does it make sense?
If you manage to end successfully the migration it would be really nice to have a wiki page describing the steps you took for this task.
Indeed!
Best, -- Didi
----- Original Message -----
From: "Martin Sivak" <msivak@redhat.com> To: "Yedidyah Bar David" <didi@redhat.com> Cc: "Sandro Bonazzola" <sbonazzo@redhat.com>, "Kostyrev Aleksandr" <kostyrev@tutu.ru>, "Users" <users@ovirt.org>, "Fabian Deutsch" <fabiand@redhat.com> Sent: Monday, March 30, 2015 11:44:39 AM Subject: Re: [ovirt-users] Change network of ovirt-nodes of a self-hosted engine cluster
Hi,
the steps make good sense indeed, but use global maintenance mode as Sandro suggested.
Good luck and keep us posted.
-- Martin Sivák msivak@redhat.com Red Hat Czech RHEV-M SLA / Brno, CZ
----- Original Message -----
----- Original Message -----
From: "Sandro Bonazzola" <sbonazzo@redhat.com> To: "Kostyrev Aleksandr" <kostyrev@tutu.ru>, "Users" <users@ovirt.org>, "Martin Sivak" <msivak@redhat.com>, "Yedidyah Bar David" <didi@redhat.com>, "Fabian Deutsch" <fabiand@redhat.com> Sent: Monday, March 30, 2015 9:44:05 AM Subject: Re: [ovirt-users] Change network of ovirt-nodes of a self-hosted engine cluster
Il 30/03/2015 08:40, Sandro Bonazzola ha scritto:
Il 25/03/2015 14:32, Kostyrev Aleksandr ha scritto:
Good day!
I'm in position when I have to change network settings of three ovirt-nodes that comprise my cluster with hosted engine. Am I correct that to change network is not a big deal and it is possible? My plan was: 1. shutdown all vms (except engine itself) on all nodes. 2. enable maintenance mode on all nodes, except the one with running engine
I think you should use Hosted Engine global maintenance while doing this. If for some reason the gateway configured in hosted_engine.conf become unreachable I'm not sure it won't try to migrate the VM. Adding Martin to the loop.
Also if for some reason the storage become unavailable during the network settings update, the engine VM will be paused due to missing storage. I'm not sure about what will happen with the SSL certificates, didi?
In 3.5+, the certs all use the hostnames, which I understand are not intended to be changed, so no issue there.
That would actually be 3.5.1+.
If some of the hosts were deployed with <= 3.4 and upgraded, the certs will use the IP address, and this will not change unless you manually reinstall (re-deploy?). Didn't try that. I guess at least some things will fail.
3. change network settings in engine node 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other 6. start engine with hosted-engine --vm-start 7. change VLAN tag at Logical Network for VMs 8. activate nodes 9. start vms
Does it make sense?
If you manage to end successfully the migration it would be really nice to have a wiki page describing the steps you took for this task.
Indeed!
Best, -- Didi
-- Didi
.... 4. shut engine down with hosted-engine --vm-stop 5. change network settings on all nodes, verify that nodes can ping each other what about step 5? How do you manage it if engine is down? Manually on all nodes? And then what if you have persistence of network config in place? Gianluca
participants (5)
-
Gianluca Cecchi -
Kostyrev Aleksandr -
Martin Sivak -
Sandro Bonazzola -
Yedidyah Bar David