5520 + 8600 problems 1

[pix530]

We have 2x8600 in the $core and 5520(ver 5.0.7.017) as access layer(stacked x2). So for some switches observe strange behavior - 1 of 2 uplinks is working, another is not.

we checked fiber, gbics.

So it seems the problem is with STP, all looks good from the configuration point of view.

And yesterday, we made some changes on the core trying to find the problem, after we lost both links connections between 5520 and 8600, but...leds on working connection are the same - seems good.
SMLT is set to normal.

And we have loose of ping every 4 is ok, then loose.

 

[DaddyOfThree]

Well there could be any number of issues.

It's probably best to try and start slowly. I would start with disconnecting one of the SMLT links and see if your situation stabilizes. You do realize that you MUST disable STP over SMLT links right? I would confirm that you have STP disabled on both the ERS8600 side and the ERS5520 side. Have you defined the ERS5520 side as a MLT? You haven't enabled VLACP have you?

Have a look at this post;

http://www.tek-tips.com/viewthread.cfm?qid=1469427&page=1

Halfway down the page I documented on to configure a basic MLT uplink using two ports.

Good Luck!

 

[pix530]

Thank you for your answer!

I am going to check all about STP in network, I had no idea about STP and MLT conflicts.

Yes, for MLT i have the following:

! *** MLT ***
no mlt
mlt 1 name "T#1" enable member 1/48,2/48 learning normal
mlt 1 learning normal
mlt spanning-tree 1 stp 1 learning normal
mlt 1 bpdu all-ports
mlt 2 name "T#2" disable member NONE learning normal
mlt 2 learning normal
mlt 2 bpdu all-ports

what do you think about this config?

VLACP - I have on core 8600:
config vlacp enable

 

[DaddyOfThree]

You've posted some good information...

It looks like you have STP enabled with the following statement;
"mlt 1 learning normal"

You should disable STP at the MLT and port level on both sides (ERS5500 and ERS8600). You can use Device Manager or you can use the CLI interface. Once you disable it at the port level it should be automatically disabled within the MLT and of the other ports in the MLT.

If you have VLACP enabled on the ERS8600 did you enable it on the ERS5500? If you haven't enabled VLACP on the ERS5500 that's probably what's causing your ports to literally bounce up and down (as VLACP keeps trying to restart the port since it can't establish with the ERS5500 switch).

You can use the following commands to enable VLACP on the ERS5500 series (assuming your at v5.x or later);

5520-48T-PWR (config)# interface fastEthernet 47,48
5520-48T-PWR (config-if)# vlacp port 47,48 timeout short
5520-48T-PWR (config-if)# vlacp port 47,48 enable
5520-48T-PWR (config-if)# exit
5520-48T-PWR (config)# vlacp enable

After that is done check the logs on your ERS8600 with the following;
ERS-8600:5# show log file tail

Also on the ERS5500 try the following commands;
ERS-5520#show interfaces 47,48
ERS-5520#show vlacp interface 47,48

Let us know how you make out.

Good Luck!

 

[pix530]

Thank you Sir for your answer, it really could help me.

I still cant understand in this copfig:

no mlt
mlt 1 name "Trunk #1" enable member 1/48,2/48 learning normal
mlt 1 learning normal
mlt spanning-tree 1 stp 1 learning normal
mlt 1 bpdu all-ports

2 and 3 commands belongs to MLT learning or STP? Why I ask you - because i menshioned 4th command, and i guess its about STP. I wrote - mlt spanning-tree 1 stp 1 learning disable. And now some swithches work with learning disable well, some with learning normal.

"You should disable STP at the MLT and port level on both sides (ERS5500 and ERS8600). You can use Device Manager or you can use the CLI interface."
- Can you provide me with more info about it - how to do. I disabled STP on FE links and on MLT nothing changed. How to disable on core? And I cant find how to do it using DM.

How I am going to play with VLACP

And question about software version - I have different versions and like to do upgrades for all switches - which versions for 5520 and 8600 are stable and recommended?

 

[DaddyOfThree]

I think I see your problem here... It looks like you're using the ASCII configuration generator "show running-config"?

I can understand why you are confused... here's the MLT section on one of my switches;

no mlt
mlt 1 name "Trunk #1" enable member 47-48 learning normal
mlt 1 learning normal
mlt 1 loadbalance basic

Here's the STP section on the same switch;

spanning-tree stp 1 add-vlan 1
interface FastEthernet ALL
spanning-tree port 1-44 learning fast
spanning-tree bpdu-filtering port ALL timeout 120
no spanning-tree bpdu-filtering port ALL enable
exit
interface FastEthernet ALL
spanning-tree port 47-48 learning disable
exit

So while it's enabled in the MLT area it's disabled in the STP area. I rarely ever use the "show running-config" command I rely on the show and config commands.

5520-48T-PWR #show spanning-tree port 47-48
Port Trunk Participation Priority Path Cost State
---- ----- --------------- -------- --------- ----------
47 1 Disabled
48 1 Disabled
sw-5520-sp2.mdc.mlhs.org#

From the ERS8600 switch you can check with the following command;

ERS8600:5# show port info stg main port <CARD/PORT>

What I was trying to say was that if you disable Spanning Tree at the port level the MLT configuration will adopt that setting and it will be automatically disabled within the MLT.

Cheers!

 

[pix530]

Thank you, I configured VLACP and now i have 1 link, but still instead of 2. I did as you wrote me. After I have on 5520:

#show interfaces 1/48,2/48
Flow Unit/Port Trunk Admin Oper Link LinkTrap Negotiation Speed Duplex Control
1/48 1 Enable Up Up Enabled Enabled 1000Mbps Full Asymm
2/48 1 Enable Down Down Enabled Enabled

#show vlacp interface 1/48,2/48
===============================================================================
VLACP Information
===============================================================================
PORT ADMIN OPER HAVE FAST SLOW TIMEOUT TIMEOUT ETH MAC
ENABLED ENABLED PARTNER TIME TIME TYPE SCALE TYPE ADDRESS
-------------------------------------------------------------------------------
1/48 true true no 500 30000 short 3 8103 01:80:c2:00:11:00
2/48 true false no 500 30000 short 3 8103 01:80:c2:00:11:00

I checked STP on the core, for my interface I have
ethernet 7/27 stg 1 stp disable.

Anyway 1 link only is working (

 

[DaddyOfThree]

Have you enabled and configured VLACP on the ERS8600? The VLACP timers and default settings do NOT match between the ERS 5500 and the ERS 8600, you need to reconfigure the ERS 8600 timings to match.

config ethernet <card/port> vlacp fast-periodic-time 500
config ethernet <card/port> vlacp timeout short
config ethernet <card/port> vlacp enable
config vlacp enable

I'm assuming that you have the fiber pairs properly cabled (have you tried swapping the fiber pairs). I don't believe VLACP will actual admin-down a port (you should see link).

Try the following commands on the ERS 8600;

show vlacp info
show port info vlacp port <card/port>

As I previously suggested you might consider disabled VLACP until you can get both links in the MLT up and running (take small steps - easier to troubleshoot).

Good Luck!

 

[pix530]

Have you enabled and configured VLACP on the ERS8600?
-Yes, it was enabled.

I checked both 8600 using DM and found that for uplink port it is adminenable, fastperiodictimer is 500 and timeout is short.

If i disable vlacp in general could it cause problems for other part of our production network?

First command shows that VLACP is enabled, second - admin and oper enabled=true, port state is down for one switch and admin and oper enabled=false for another.

I checked 5520 - on both devices STP is disabled and VLACP is enabled for uplinks.
Then I checked 8600 - vlacp is enabled, stp is ...forwarding for 1 switch. Then I disabled STP but - State is forwarding, Enable STP is false (...I did disable-enable port - no changes.

 

[DaddyOfThree]

I don't believe VLACP is your problem. You should have physical link between the two GBICs/SFPs regardless of VLACP.

I would suggest you look at the physical aspects. Is the fiber bad? Can you test/replace the fiber? Are the GBICs/SFPs bad? Can you replace them? You don't have the fiber pairs Tx/Rx crossed do you? You've got auto-negotiate enable on both sides?

Cheers!

 

[lobk]

Reading your post it doesn't sound like you have SMLT configured correctly on your 8600.

"Then I checked 8600 - vlacp is enabled, stp is ...forwarding for 1 switch. Then I disabled STP but - State is forwarding, Enable STP is false (...I did disable-enable port - no changes."

On the 8600, when you configure an MLT for SMLT it will disable STP on the MLT port. I don't think you configured SMLT correct if yuor saying you saw STP enable on the port from the 8600 going to the 5520 and you have it configured for SMLT.

Even if STP is your problem it should take down the link. The port should be in a block state. Look at your log.

 

[DaddyOfThree]

Did you ever figure things out pix530?

 

[pix530]

Guys, i appreciate your help!

Now after playing games with all these switches, I established 2 links with 2 switches, so...couple of things, last one was smlt was not created on core.

Now I wait for cable guys to test other 3 switches, I checked configs and seems ok, the only thing i found - some switches work fine with vlacp and some without enabled.

I switched off STP everywhere on MLT links.

 

[DaddyOfThree]

Thanks for the reply... glad to hear your moving forward.

One thing I've noticed with VLACP on the ES460, ES470, ERS5500 series is that sometimes you need to reset the switch to get VLACP fully functional. I've seem numerous times where I enable VLACP and it think the peer is up when I haven't even configured the peer yet. I quick reset of the switch and it shows the peer as down (until I configure the ERS8600 on the other side).

Good Luck!

 

[pix530]

Guys, sad to say. One morning second link was down...

So...After some attempts to establish situation I mentioned the following:
We have 2x8610(4.1.4) in the core and 5520(ver 5.0.7.017) as access layer(stacked x2).

03/07 morning this switch 5520 comes down from 1 working link. From that time we cant access to our servers connected to it.

So we observe strange behavior - 1 of 2 uplinks is blinking as it is ok , another is not.(uplink on base unit is seems ok but it is not true, on slave unit is not blinking at all)
we checked fiber, gbics.
all looks good from the configuration point of view.

Strange behavior :
- from 5520 i can ping default gateway
- from core i can ping 5520
- from any other part of the network i cant ping 5520
- from any other part of the network i can ping servers connected to slave unit.
- from any other part of the network i can not ping servers connected to base unit.

our steps:
- erased configuration for 5520 and wrote it from 0 again
- disabled vlacp from both ends - the same problem
- configured another port (8/27 instead of 7/27) on the core for slave unit. It changed color in EMS view of core from orange to green, still the same behavior for ping and access.

So after all i changed ports on the core and configured new smlt - nice - both links are green, servers are accessible...But...I still can not connect to the 5520.





More, with another switch we have VERY strange behavior - I configured new stack, and I can not ping but can telnet from 1-2 attempts. Very strange (

 

[DaddyOfThree]

Hmmm...

If you have a Nortel support contract I would advise you to use it. It doesn't sound like this is a configuration issue but I'll try to throw some questions out there.

I would immediately advise you to upgrade the ERS5520 to v5.1 software. There were all sorts of issues with the 5.0 release and a few versions were even pulled from the Nortel website.

Let me fast forward to your final question.

Why can't you ping the ERS 5520 management IP address?

Let's just assume that this is a configuration issue. I can't see how it wouldn't be. First the ERS5520 switch. Are you sure you configured the "In-Band Stack IP Address" and not the "In-Band Switch IP Address" in the IP/Configuration Setup screen?

Assuming that the IP address is configured properly let's try disconnecting one of the uplinks (port 1/48). Wait 30-60 seconds and initiate your ping test. Are you successful? Reconnect that uplink (port 1/48) and wait 30-60 seconds and disconnect the second uplink (2/48)and wait 30-60 seconds and repeat your ping test. Are you successful? If you are successful you could have a SMLT/IST issue.

I would also repeat your ping test from both core ERS 8600 switches. Do both ERS 8600 switches have an IP address in the management VLAN that the ERS 5520 is configured for?

If you are using a VLAN other than the Default VLAN have you configured it on the ERS 5520 and set it to be the management VLAN?

Good Luck!

 

[pix530]

I solved all issues to the moment. Found some problems with mlt.

What i mentioned in one situation - disable-enable mlt helps.

Thank you DaddyOfThree for help and attention.

 

[rjenk]

Just a quick suggestion, did you verify proper operation of the IST between the 8600's? If you had a IST MLT between the two 8600's and for whatever reason the IST was not working, when you brought up the second links to the edge equipment you would have a loop.