One vertica node in a three node cluster is down

Products

• CA Infrastructure Management
• CA Performance Management

Issue/Introduction

How do I start start one vertica node in a three node cluster if it is determined by looking at the vertica.log that there is corruption?

Environment

Any CAPM version

Resolution

In the vertica.log you may see something like this:

v_drdata_node0002/normal/vertica.log-2017-03-17 07:56:43.959 DistCall Dispatch:0x7fc0fc00bad0-a00000003913b0 [Recover] <INFO> library file: [TxtIndex_TxtIndexLib_45036002708994452.so] is missing.

v_drdata_node0002/normal/vertica.log-2017-03-17 07:56:43.959 DistCall Dispatch:0x7fc0fc00bad0-a00000003913b0 [Recover] <INFO> library file: [public_FlexTableLib_45036002708994592.so] is missing.

v_drdata_node0002/normal/vertica.log-2017-03-17 07:56:43.959 DistCall Dispatch:0x7fc0fc00bad0-a00000003913b0 [Recover] <INFO> library file: [public_ApproximateLib_45036002708994830.so] is missing.

 

 

Messages like this would be an example of some level of corruption.  To force a repair of corruption in he database You can run the following command

 

• /opt/vertica/bin/admintools -t restart_node -s <ip_address_of_node2>  -d <db_name> -F 

After this the vertcia node should start successfully

Additional Information

A caveat to this command is that the other nodes in a 3 node cluster must be up and running or this won't work