Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

81846

April 7th, 2014 05:00

FW 7.0.2 - Synrep Issues and Errors

Dear Everyone,

After our upgrade from v6.0.4 to v7.0.2, we are getting strange errors by email and logging. These errors started after the upgrade.

Strangest thing is that when we are receiving an error and are logging in, the error is gone already.

Severity  Date and Time      Member       ID              Message                                                                                                                                                                                                                                                                  --------  -----------------  -----------  --------------  -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------   Error    7-4-2014 11:44:20  san4-shared  7.4.3 | 7.4.23  iSCSI login to target '10.x.x.x:3260, iqn.2001-05.com.equallogic:xxxx'from initiator '10.x.x.x:55898, eqlinitiatorsyncrep' failed for the following reason: | Initiator disconnected from target during login. 

 

 

Why is this error coming up so often since the upgrade? Is this a Hardware or Software failure? The reason i'm asking is that our other clusters are not showing this error.

With regards,

Andreas van der Linden.

11 Posts

April 23rd, 2014 06:00

Dear Everyone,

Because of your discussion I have taken over the function of our network engineer and did some exploring myself. We don't know why, but someone enabled spanning tree on these interfaces. We are planning a maintenance window shortly to turn this off.

Thank you very much for your suggestions.

5 Practitioner

 • 

274.2K Posts

April 7th, 2014 16:00

what is the connection between the pools?   What speed and latency?

11 Posts

April 8th, 2014 01:00

The connection between the SAN's is configured as followed:

- Single Cisco Redundant stacked switch between the SAN's

- MTU 9000

- 1-4 Gb/s

Type: 2x PS4000 and 2x PS6100 where the sync rep is only on the same type (4000 -> 4000 and 6100 -> 6100)

5 Practitioner

 • 

274.2K Posts

April 8th, 2014 09:00

Can you clarify what "1-4Gb/sec" means?  

11 Posts

April 8th, 2014 13:00

depending on the type of equallogic, it has 2x1Gb/s interfaces or 4x1Gb/s interfaces to the switch.

5 Practitioner

 • 

274.2K Posts

April 8th, 2014 13:00

So both pools are on the same switch?   What model Cisco?

11 Posts

April 9th, 2014 01:00

Cisco C3750E

11 Posts

April 14th, 2014 06:00

no one?

5 Practitioner

 • 

274.2K Posts

April 14th, 2014 09:00

I would suggest opening a support case so they can review the diags and switch logs.

There are some QoS settings for the 3750 line that improves performance but reducing retransmissions.

127 Posts

April 18th, 2014 05:00

Please check if the upgrade to v7 somehow enabled DCB if it was disabled before the upgrade. If you find it enabled, please diable it (only possible via CLI, not via GUI with v7).

Let us know if that helped.

Best regards, Joerg

5 Practitioner

 • 

274.2K Posts

April 18th, 2014 08:00

FYI:  you don't disable DCB on the EQL array side, ever.   Only do so on the physical switch.

Also DCB is only applicable to 10GbE interfaces, in this case, it's just GbE.

127 Posts

April 19th, 2014 04:00

Don: Please clarify.

Many switches will turn off any flow control if you disable dcb on the switch side. Thus - please explain.

Why am i not to disable DCB on the array side, ever? That does not make any sense to me, because a) it was possible for a long time from within the gui and b) maybe you don´t want to disable it on the switch side for several reasons but you need to disable it on the array because the initiator don´t supports it. Also, i often spoke to EQL techs who told me where to disable it on the cli side for exatly that reason. Also, it is clearly explained in the cli guide. So i am concerned now: Is there *anything* we should know about that and why do you explicitly mention it?

I had a coincidental chat about that with EQL techs in frankfurt/germany last week during the dell enterprise forum and it was exactly about dcb and how dcb is not widely supported by initiators, especially s/w initiators and how to disable it on the array level.

Don maybe you remember this we had a conversation in this forum about a year ago on exactly the same topic - but you never told me why it is so important to you to explicitly not disable it an the array level?

I really want to know the reason because if there is something important i should know about that i´d like to understand that and disable it on the switch side instead of the array side. But as far as i know and also as far as my friends and colleagues in frankfurt are telling me, disabling dcb on the array side does no harm at all. Please clarify.

Best regards,

Joerg

5 Practitioner

 • 

274.2K Posts

April 21st, 2014 11:00

Re:  DCB and flowcontrol.     With DCB enabled on a switch, Link Layer Flowcontrol is disabled, since Priority Flow Control is preferred.  PFC only works when DCB is enabled, configured and negotiated properly.  So turning OFF DCB on the switch allows Link Layer flowcontrol to be enabled and negotiated on the switch.

I received the statement about not turning it off on the array side directly from engineering.

127 Posts

April 22nd, 2014 02:00

Hi Don,

OK, thanks. Could you talk to engineering for the reason for telling you this? Please do me the favor because the enginners i myself talked to, told me otherwise.

Second, please give exact advise how to disable DCB on a DELL 8164 (now N4064) so that it´s really disabled while in parallel keep link layer flow control *fully* intact. Because i don´t know how to realize that.

Best regards,

Joerg

5 Practitioner

 • 

274.2K Posts

April 22nd, 2014 10:00

From the Best Practices guide for DCB + EQL   Page 26.

en.community.dell.com/.../4355.configuring-dcb-with-equallogic-sans.aspx

Network configuration assumptions:

• All switches in the Layer 2 network carrying iSCSI traffic must be configured to be either DCB

enabled or DCB disabled. Mixed configuration with a partial set of switches in non-DCB mode

should not exist in the Layer 2 DCB network domain. The administrator must ensure that all switches

are either in DCB enabled state or DCB disabled state on the network during the configuration

process.

Part of the IEEE DCB standard says that all end nodes must always advertise "willing" mode.

A comment from one of the EQL designers might help here.

"disabling" DCB at the end system doesn't do anything helpful.  If

the switch is doing the wrong thing, the fact that the EQL array isn't paying

attention to DCB doesn't help at all.  DCB is *not* a negotiation, it's a unilateral

choice made by the switch.

I'll check on the 8164.  [tag:no] dcb enable is what I recall.   I'll double check and make sure there's not another command to make sure LL flowcontrol is enabled.

No Events found!

Top