CLI Guide

Table Of Contents

● <tla>-<cluster>-diag-ext-<datetime>.tar.gz- Contains java heap dump, fast trace dump, two latest core files,

and two latest core files ( if --noextended is used, then no core file is collected).

● <tla>-<cluster>-diag-<datetime>.tar.gz - Contains everything else including a direcotry, /opt/recoverpoint

which contains the RecoverPoint splitter logs in a .zip file (vpsplitter.log.xx, vpsplitter.log.periodic_env, and

vpsplitter.log.current_env).

NOTE: If --noextended is used, then this file is not available.

NOTE: The collect-diagnostics must be run as the service user to collect all the available files. For this, the VPlexcli

command must be invoked as the service user.

Best practice is to collect both files. The extended file is large, and thus takes some time to transfer.

Recommended practice is to transfer the base file (tla-diagnostics-timestamp.tar.gz) first and begin analysis while

waiting for the extended file to transfer.

NOTE: On metro node Metro configurations, collect-diagnostics must be invoked from each management server in order to

collect complete diagnostics. The management server diagnostics files are only collected for the local cluster.

The director diagnostics are retrieved from ALL directors in a metro node Metro unless the --local-only argument is used.

All trace files under the folder /cores/nsfw-trace/ on each director will be copied and put in the collected diagnostics.

CAUTION: In metro node Metro configurations, run the collect-diagnostics command on each management

server, but NOT at the same time. Even if you use the --local-only argument, do not run the command on

both management servers at the same time.

If you know the time period in which an event happened, you can collect a subset of logs based on time period. This reduces the

time to collect diagnostics. Use the --log-range and --last-logs arguments to collect a subset of diagnostics.

Remove files created by collect-diagnostics from the management server as soon as possible to avoid filling management server

disk partitions.

Example

Collect diagnostics for the entire cluster:

VPlexcli:/> collect-diagnostics

If this is a metro configuration, then run it on each cluster independently.

VPlexcli:/> collect-diagnostics --noextended

Collect diagnostics, omitting trace files on the directors and the management server console heap, and send the output to the

default directory.