Basic Questions

Question

Hi there,

I have number of questions some are very basic and some need expert opinion. I ll explain the config we have and will ask all my questions at the end.

During data mover failover, when the standby boots up does it use the physical ports that were connected to primary?

I am trying to understand how many physical ports should be active for following configuration.

One fsn device (cge0 and cge1 of primary DM)
cge3 and cge4 (both of primary DM) are assigned to single iSCSI target that has number of LUNs defined to it.

I understand it as follow

Only 4 network cables should be connected between Primary DM and the network (two diffrent VLANs, one for fsn and other for iSCSI target). And i should still have 4 physical ports available (at Standby DM) that can be used for other purposes. my understanding is that when data mover fails over it takes over Primary DMs identity and uses all the slots assigned to primary DM. Please correct me if i am missing any IP addresses assigned to the interfaces.......

Is there any load balancing between the two ip addresses assigned to the network portals of the iSCSI target? Is it same as fsn?

Any cautions before deleting a CIFS server that has no shares on it?

Can Celerra 5.5 be upgraded to 5.6 easily? Is there any downtime involved in the upgrade?

Any good documentation if i want to add another DAE to integrated NS20? I will open service call for this but just to get familiar with the process.

Thanks alot for taking the time to read this and I really appreciate the responses.

nandas · Accepted Answer

Hi,It would have been better to have separate posts for each question. Anyway - I'll try to provide some details from my end and I am sure other experts in this forum are also going to help you.During data mover failover, when the standby boots up does it use the physical ports that were connected to primary?The standby data mover does not use the physical ports connected to primary data mover. Seperate but identical physical connection must be there on the standby data mover, in order to have proper failover of data mover. In other words, you need to have same physical connection and network switch configuration for both primary and standby data movers.Only 4 network cables should be connected between Primary DM and the network (two diffrent VLANs, one for fsn and other for iSCSI target). And i should still have 4 physical ports available (at Standby DM) that can be used for other purposes. my understanding is that when data mover fails over it takes over Primary DMs identity and uses all the slots assigned to primary DM. Please correct me if i am missing any IP addresses assigned to the interfaces.......As I mentioned earlier, you need to have 4 network cables for Primary data mover and 4 cables for standby data mover. The switch ports to which cge0 and cge1 of the standby data mover are connected needs to be configured in the same VLAN as the FSN. Similarly the switch ports to which cge2 and cge3 of standby data mover are connected needs to be in the iSCSI VLAN. When the data mover fails over, the standby data mover takes over the identity of the primary data mover, but the physical connectivity has to be there and has to be the same for the standby data mover to work properly. The identity is the virtual - in normal case, say for example slot 2 is working as server_2 (Primary) and slot 3 is working as server_3 (standby). when the data mover 2 fails over to data mover 3, slot 3 gets renamed as server_2 thus taking over the identity of primary data mover - the IP Configuration, File systems, shares etc (everything configured for server_2) are now available through slot_3 which is working as server_2 in failed over case. So, for standby data mover, you need only the physical connection and identical switch port configuration - but no IP address is required.Is there any load balancing between the two ip addresses assigned to the network portals of the iSCSI target? Is it same as fsn?Yes - you may configure either EtherChannel trunk or LACP (Link aggregation Channel protocol - IEEE 802.3ad standard) trunk on these two ports - and assign the IP address on the virtual interface of the trunk - not the individual cge2 or cge3 interface. In this case, both the cge2 and cge3 ports will work as a virtual team (channel). If you plan on doing so, please ensure that the respective ports for standby data mover are also configured in a channel on the network switch end.Please refer to the following documents in powerlink - which will provide more clarification -'Configuring and Managing Celerra Network High Availability''Configuring Standbys on Celerra'Any cautions before deleting a CIFS server that has no shares on it?You may delete the CIFS server if there is no shares on it - but please ensure that the CIFS server is not used for any other purpose - say for example CAVA (Celerra Antivirus Agent) scanning, or NDMP backup or any other management purpose etc. However, leaving the CIFS server as it is will also not cause any issue I guess.Can Celerra 5.5 be upgraded to 5.6 easily? Is there any downtime involved in the upgrade?Yes - it can be upgraded - proper upgrade path is available. Upgrade complexity may depend on existing configuration and what all features are in use now. It requires a downtime - please get in touch with your EMC Customer Engineer or other EMC Representative for more discussion on this.Any good documentation if i want to add another DAE to integrated NS20? I will open service call for this but just to get familiar with the process.Since this is an integrated NS20, you may simply run the setup_clariion command from CLI or from Celerra Web Manager GUI -> Go to Storage -> Systems -> right click on the array and choose configure. Please ensure you have proper number of support disks and choose the proper template. You may take a look at the following -http://corpusweb130.corp.emc.com/ns/common/post_install/Config_storage_integrated.htmRequest others to provide valuable suggestions. Please feel free to revert back, if required.Thanks,Sandip

YDroid · Answer

Sandip,Thanks for the answers things make more sense now, I need a little bit of clarification on two things out of your answers1. when you mention slot2 and slot3 does it mean that slot2 is for whole blade? Reason i am asking is that i see slots when i run following commandserver_sysconfig server_2 -pciserver_2 : PCI DEVICES:On Board: VendorID=0x10B5 DeviceID=0x8516 Controller 0: PLX PEX8516 IRQ: 11Slot: 1 Agilent QX4 Fibre Channel Controller 0: fcp-0 IRQ: 24 1: fcp-1 IRQ: 25 2: fcp-2 IRQ: 26 3: fcp-3 IRQ: 27 Slot: 3 Broadcom Gigabit Ethernet Controller 0: cge0 IRQ: 22 1: cge1 IRQ: 23Slot: 2 Broadcom Gigabit Ethernet Controller 0: cge2 IRQ: 20 1: cge3 IRQ: 21Are these slots just for the pci cards and diffrent from slots you are referring to in the answer? Also what is that OnBoard device? Is it the two network cards connecting blade to the control station? Do they appear as mge0 and mge1 with ifconfig command? 2. How can i check if my global CIFS server is used for CAVA or NDMP (we have both CAVA and NDMP configured)

YDroid · Answer

Thanks for clarification Rainer.Can you answer my second question about how to see what my global CIFS server with no CIFS shares is configured for.thanks

Rainer_EMC · Answer

Only 4 network cables should be connected between Primary DM and the network (two diffrent VLANs, one
for fsn and other for iSCSI target). And i should still have 4 physical ports available (at Standby DM)
that can be used for other purposes.

what do you mean with "available for other purposes"

To be clear - a standby DM doesnt do anything until there is a datamover failover.
Then it takes over the IP, MAC address and all the file systems and services of the primary DM

Rainer_EMC · Answer

sorry - thats a bit confusinghistorically the data movers were in slots in the old machines (CNS/CFS14) - thats why the first data mover is called server_2 (slot 0 and 1 were the controlstations.These days there arent physical slots but the software still uses them logical.What you are listing as slots with server_sysconfig are slots within the data mover where the network and fibre channel cards are located

YDroid · Answer

I got this one clear about the network ports on standby datamover.

Now my question is diffrent, its about a CIFS server that i have configured my EMC Implementation specialist. This CIFS server is created on primary datamover. There isnt any share created on it at all. I was told by Sandip that sometimes a global CIFS server is created for CAVA and/or NDMP. How can i check what is the purpose of this CIFS server.

also if you can tell me what is (this appears by running server_sysconfig command for primary datamover

On Board:
VendorID=0x10B5 DeviceID=0x8516 Controller
0: PLX PEX8516 IRQ: 11

thanks

nandas · Answer

Sorry - I could not login to Forum for sometime thus could not answer your queries. Thanks to Rainer to help clarifying your doubts and answering questions.

I see you have a new thread for CIFS server with no share question - Let me put my comments in that thread.

Cheers,
Sandip

Celerra

Basic Questions

Was this post helpful?