Central Manager failover process

Unanswered Question
Aug 12th, 2010

Testing the CM primary/standby process.

WaaS software version 4.2.3

In my lab:

WAVE-574 primary CM

WAVE-474 standy CM  

WAE-674 app accelerator

failover test scenario:

reboot the primary CM

watch the CMS info status on the 674.  Nothing really changes until I issue the command "central manager role primary" on the standby unit (as expected)

odd behavior:

as the former primary unit goes down from the reboot - GUI IP address is no longer reachable at the primary IP.  Makes sense..the device went down hard.  (is there no sync of IP address between the CM's?)

primary CM is now offline.  standby CM is alive...but needs to be promoted - so i do this via the CLI on the standby unit.

Now - back to the GUI.  The only way to continue to manage the environment is to reach the GUI of the standy unit...seems like a pretty lame failover process...you have to open a new browser session and make a connection to the IP of the standby unit.

More funkyness:

As the primary comes back online - it comes back as standby - which is cool - somehow it knew the other unit was primary - but now neither are functioning.  Somehow the standby unit automatically went back to standby status, but when I do a show run:

central-manager role halted

WAVE574 primary unit:


cms-primary#show cms info
Device registration information :
Device Id                            = 166                                
Device registered as                 = WAAS Central Manager               
Current WAAS Central Manager role    = Standby                            
Current WAAS Central Manager         = 139.126.176.216                    
Registered with WAAS Central Manager = 139.126.176.216                   

CMS services information :
Service cms_httpd is not running
Service cms_cdm is not running
cms-primary#

WAVE474 standby unit (that was promoted manually)

cms-standby#show cms info
Device registration information :
Device Id                            = 404                                
Device registered as                 = WAAS Central Manager               
Current WAAS Central Manager role    = Standby                            
Current WAAS Central Manager         = 139.126.176.216                    
Registered with WAAS Central Manager = 139.126.176.216                    
Status                               = Offline                            
Time of last config-sync             = Thu Aug 12 16:40:04 2010          

CMS services information :
Service cms_httpd is not running
Service cms_cdm is not running
cms-standby#

neither unit are reachable via the GUI interface.

WAE674 app accelerator status:

mpls-lab-wae674#show cms info
Device registration information :
Device Id                            = 1106                               
Device registered as                 = WAAS Application Engine            
Current WAAS Central Manager         = 139.126.176.217    <-- standy device                
Registered with WAAS Central Manager = 139.126.176.216  <-- primary device                    
Status                               = Offline                            
Time of last config-sync             = Wed Aug 11 13:46:53 2010          

CMS services information :
Service cms_ce is running
mpls-lab-wae674#

Is this really the expected behavior?  The documentation on this whole thing pretty much just shows you how to configure the devices, but I have yet to find a detailed document about how this works behind the scenes...so i'm not sure if i'm doing something wrong here, or this is just how it is?  Could use some feedback from the community.

Thanks,
Paul

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
wrobbin Thu, 08/12/2010 - 10:40

That is odd...im actually turning up a standy Central Manager next week any advice

seem really easy and strait foward... Kinda looks like when it came back up it did a configsync.

pdinapoli Thu, 08/12/2010 - 10:54

i would agree - the configuration is very simple, but the funcationality is pretty poor in my mind.

It looks to me that the process of promoting the standby to primary works fine - you'd just have to know that your CM GUI is now reachable at the standby unit IP not the primary - so its not like the standy unit takes over the IP of the primary.  this is a support problem for large organization where the 1st or 2nd tier team might not no unless we tell them " your CM GUI IP is now reachable via standby IP of a.b.c.d"

Gets more interesting:

as the primary unit comes back online - it sees that there is alreayd a primary - so it switches itself to standby.

with no intervention on the standby unit (the actual standby WAVE474) at some point it switched itself to standby.  as the primary unit came back online was also standby - so neither were functioning.

doing a show run revealed that both had this line in their CLI

"central manager role halted"


i had to manually restore primary back to primary, standby to standby and I had to say "cms enbable" on both to get it all back to nornal.

kapsabha Thu, 08/12/2010 - 12:18

The following link mentions the CM switch over procedure:
http://www.cisco.com/en/US/docs/app_ntwk_services/waas/waas/v421/configuration/guide/maint.html#wp1159476

The Abstract from the above link in the configuration Guide for WAAS suggests that

If you switch a warm standby Central Manager to primary while your primary Central Manager is still online and active, both Central Managers detect each other, automatically shut themselves down, and disable management services. The Central Managers are switched to halted, which is automatically saved in flash memory. Hence is an expected behaviour.

Thanks and Regards,
Kapil Sabharwal

WAAS PDI Help Desk
www.cisco.com/go/pdihelpdesk

Actions

This Discussion