Cluster Master Line Commands (continues)

Cluster Master Line Commands (continues)

Product Line:  VERDE

 

ISSUE: 


The following are commands that are associated with the control or information of the Cluster Master (CM)

How to verify which VERDE server is the Cluster Master?

From  any VERDE server, type "verde-util -am-i-it" it will display which server is the Cluster Master.
(starting from VERDE 6.6, the new syntax is LD_LIBRARY_PATH=/usr/lib/verde/etc/libs /usr/lib/verde/bin/verde-util --am-i-it )

When a count continues to display in the Management Console's RESERVE column, a forced reset must be done by demoting the CM and forcing a CM fail-over.  This will force a clear of the counts in the RESERVE column:

The command to demote the Cluster Master:

Run the command /usr/lib/verde/bin/verde-util --demote

NOTE: This is done assuming the other node is setup as CM + VDI.

The command to promote a VERDE VDI server to a Cluster Master:

Run the command /usr/lib/verde/bin/verde-util --promote     

or

/usr/lib/verde/bin/verde-util --force --promote (To force the server to a Cluster Master)

The command to do see if health check is implemented:

Run the command /usr/lib/verde/bin/verde-util --hc
returns 0 if health check successful (i.e. $MC_USER/.verde/dbaddress is fresh)
returns > 0 otherwise

If there are multiple Cluster Masters and to change one CM that is not active to the active CM, do the following:

In a cluster with 2 CM nodes, say CM1 and CM2. CM1 is active.  By demoting CM1, now both CM1 and CM2 race to become the active CM and there's a 50% chance CM1 will become active again. the right way to promote CM2 to active is to perform a "verde-util --force-promote" on CM2.

To check if by chance some other environment's Cluster Master has control and causing your CM not to come up perform the following:

stat /home/vb-verde/.verde-local/dbaddress

Result:
File: `/home/vb-verde/.verde-local/dbaddress'
Size: 76 Blocks: 1 IO Block: 1048576 regular file
Device: 15h/21d Inode: 12041945979643538187 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 501/vb-verde) Gid: ( 501/vb-verde)
Access: 2015-06-03 22:06:14.000000000 +0200
Modify: 2015-06-03 22:06:14.000000000 +0200
Change: 2015-06-03 22:06:14.135685564 +0200


How to set a server to Offline or an Offline server to Online:

Server is ONLINE.  Do the following to take it OFFLINE:

verdecmon-ni offlineserver server-id


The server is OFFLINE. Do the following to take it to ONLINE:
verdecmon-ni onlineserver server-id

"server-id" - VERDE_SERVER_UUID from '/VAR/LIB/VERDE/settings.node' file." 

    • Related Articles

    • Cluster Master Line Commands

      Product Line:  VERDE VERDE version: 5.5 and above Host/Server: All Guest/Image:N/A Client/Workstation: N/A   ISSUE:  The following are commands that are associated with the control or information of the Cluster Master (CM) How to verify which VERDE ...

    • VERDE Cluster Master Environment Explained

      Product Line:  VERDE Having a multi-server cluster requires a shared storage (NAS) device.  All pertinent data is stored on the NAS.  Database, gold image, user information, etc.  In a clustered env, the only thing, by default, that is stored on the ...

    • Testing and Results of Cluster Master Shutdown as well as Tomcat

      Product Line:  VERDE Testing was performed on a Clustered environment in order to obtain true behavior as a result of a Cluster Master failure and failover. Given the following configuration:   VERDE Cluster Configuration: 1 server dedicated to ...

    • Cluster Master FailOver Licensing Issue and Workaround (8.2..fixed in 8.3)

      Product Line:  VERDE The Cluster Master is used to redeem a license key with the Management Portal. Upon successful redemption the Management Portal sends licensing information (e.g., number of seats, type, duration, etc.) to the Cluster Master. The ...

    • Normal, Clean Shutdown and Start Up of a VERDE Clustered Environment

      Product Line:  VERDE Obviously, there will be misfortunate times when servers will "crash" resulting in loss of a user's connectivity and possibly their USER data.  Given the option, it's much better to perform clean shutdowns and startups of the ...