cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements
Field Notice 70545
11674
Views
0
Helpful
32
Replies
Ron Baduach
Cisco Employee

UCSM failed login , not available on secondary node

Hi Guys , hope you are all fine .

I'm getting some error when trying to login to UCS Manager [see screenshot below]

I'm familiar with this error when the primary FI is down , but here , both FIs are up [pingable] .

Can someone please assist ? we can't access UCSM , and we really need it now ...

Thanks in advance !

32 REPLIES 32

Hi Ron

Something must be really sick; this cluster lead command takes only a few seconds !

Q. was this configuration ever running properly ?

Can you please post the IP address of FI-A, FI-B and VIP ? 

Are they in the same IP subnet / vlan as the KVM addresses ?

Regarding NTP/DNS 

- did you configure DNS ? any typo ?

- did you configure NTP ? any typo ? I see that your date / time is totally wrong !

- assuming IP connectivity is ok, I would try NTP IP address instead of DNS name.

Walter.

Hi Wlater ,

1. From what i know the config ran properly .

2. FI-A - 10.56.27.19

FI-B - 10.56.27.20

FI-VIP - 10.56.20.10

the same subnet/vlan as the KVM .

3. Me personally didn't configured NTP/DNS , but the guy that worked on it , that now fired , maybe did it .

Can you dell me how to check ? and I'll configure NTP & DNS with IP

4. BTW - is there any command that can stop the switchover ?... [see screenshot]

[I tried to get back FI-A to be primary (by rebooting FI-B , and firce FI-A to be primary ...)]

Walter , thanks for your help

Hi Ron

Looks strange to me

FI-A - 10.56.27.19

FI-B - 10.56.27.20

FI-VIP - 10.56.20.10  !!!!!!!!!

why not eg.

FI-A - 10.56.27.19

FI-B - 10.56.27.20

FI-VIP - 10.56.27.21

what is the network mask ?

sorry sorry sorry

TYPO :)

VIP IP - 10.56.27.10

all of it /24

||||||||||||||||||||||||||||

BTW - now , when A is primary again , FI-A falls once in a few minutes ..

You mean FI-A crashes ?

Please check if you you see dump files !

1. yes, FI-A ,it crashed twice , but i don't think that rebooted , because in 20-30 seconds it back to life , so probably was some network connectivity down .

Now , after configure the NTP as needed [i saw that what was configured is not in use anymore...] , it's not crashed ..

2.don't know what is dump files and how to check if they are existing...

3. So now it not crashes , but i'm still with the issue that i mentioned in the screenshot ...

A middle conclusion -

In the beggining , FI-A was primary but i could not access UCSM

Now, FI-A again is the primary [after moved it to FI-B] , and now UCSM is accessible , but , i got 2 things :

a. the one in screen shot "FI-A"

b. the one in screenshot "EthPorts"

again , thanks for the help , and for keep assist with it !

In the attachments, you see how to configure DNS, NTP and check for core (dump) files.

Thanks ! Like i said in the previous comment , i succeeded with it !

1. about core [dumps] files , see the attachment .

2.Please take a look on my previous comment if you missed it

thanks again !

It seems that your time is not up to date ! NTP doesn't seem to work ! In show cluster ... I see December 2015 !

Q. is this a productive environment or lab ?

It seems that FI-B has a problem, therefore cluster lead b doesn't work.

One could disconnect the heartbeat cables, then erase configuration on FI-B, reconnect the cable, and join the cluster (entering the FI-B IP address) and then synch the configuration from FI-A.

Walter.

Hi Walter ,

The 2015 is the "start time" , but take a look at "last election time" , it's now , 2016 .

also , when running "show clock" on both FIs , the time is fine [although FI-A is in 40 minutes late , although they both have the same NTP server...]

It's between lab to production , we can get it down if needed [but , the ESXis will be down too ???]

Hi Ron 

The 2015 is the "start time" , but take a look at "last election time" , it's now , 2016 .

yes, you are of course right !

also , when running "show clock" on both FIs , the time is fine [although FI-A is in 40 minutes late , although they both have the same NTP server...]

Something with NTP is still wrong ! or did you manually configure date/time on each FI ? is F999616 still there ?

It's between lab to production , we can get it down if needed [but , the ESXis will be down too ???]

Did you check, that ESXi switch has connectivity to both fabrics ?

If yes, bringing down one fabric should not disrupt ESXi.

Walter.

PS. the core files are old ! Is this a rather old hardware ? FI 61xx ?

The 2015 is the "start time" , but take a look at "last election time" , it's now , 2016 .

yes, you are of course right !

Yay!

also , when running "show clock" on both FIs , the time is fine [although FI-A is in 40 minutes late , although they both have the same NTP server...]

Something with NTP is still wrong ! is F999616 still there ?

Nope , not there.

It's between lab to production , we can get it down if needed [but , the ESXis will be down too ???]

Did you check, that ESXi switch has connectivity to both fabrics ?

If yes, bringing down one fabric should not disrupt ESXi.

I'll check , thanks for the info !

Hi Walter , we'll probably will go on your suggestion .

"One could disconnect the heartbeat cables, then erase configuration on FI-B, reconnect the cable, and join the cluster (entering the FI-B IP address) and then synch the configuration from FI-A."

Just two things :

1. what the exact command to erase the configuration from FI-B ?

2. How do I sync the config from FI-A to FI-B ? it should happen automatically after join FI-B to the domain ? (entering FI-B IP Address ?)

Hi Ron

Enter “connect local-mgmt”

Enter “erase configuration”

--------------------------------------

During boot:

connect the heart beat cable => FI-B will recognize a peer and asks if it should join to create cluster => yes

--------------------------------------

the synch is completely automatic No action required ! just be patient ! depending on the size of the configuration it may take time 10 + minutes

thanks a lot ! will let you know once done !

Create
Recognize Your Peers
Content for Community-Ad