Support Diagnostic 6.4.2

1. Fixed issue with noVerify parameter
2. Better messaging when GH could not be accessed for version check.
3. Modified check for local node to use only IP's
4. Modified standard type value to clarify help message.

Support Diagnostic 6.4.1

Fixed issue where having http disabled on some nodes could cause log collection to be bypassed.

Support Diagnostic 6.4

New for this release:

• Version check - the utility will now check Github for a later release, and if you are not running that will present a warning along with the URL to the current release download.
• New system information outputs have been added containing system and hardware information. They are named system-digest.txt and system-digest.json. These contain an aggregation of values from some of the other system calls but also some others not previously collected such as mounts.
• New type: local. This will only collect logs and run OS specific system calls on the host the utility is run on. You can supply a log directory if the default is not being used, or if it's not known you can specify any running node in the cluster and it will attempt to find a valid log directory on that host using settings from other running nodes. There will be a set of command prompts that will walk through the process.
• An sha1 checksum is now produced.
• If multiple nodes are run on a host, logs will only be captured for the first one found.
• Related to the above captured logs are contained in a /logs directory rather than nesting under a node name folder.
• Shard level information is now captured in indices_stats.
• The type logstash now collects the system calls as well as the logstash API calls. Thanks, Jared.
• There is a new version of the scrub functionality. Rather than running as part of the diagnostic, this is now a separate application. Rather than just process the logs, it will go through every file in the diagnostic archive. By default, ip addresses (v4 and v6) will be obfuscated, as well as MAC addresses. String literal and regex's can be added to the scub,yml file, and all strings found in a search will be replaced with a generated token. Consistency will be maintained throughout the run, so if a string is encountered 5 times, all 5 times will be replaced with the same value.

Running this will be similar to running the diagnostic. Once a diagnostic has been produced, it can be run using:

./scrub.sh -a <full path to the previous archive>

This will extract each file in the archive, process it, and write it out to a temp directory. Once that is completed a revised archive with "scrubbed-" prepended to the name will be written out to the same directory that contains the original archive and the temp directory will be deleted.

Fixed

• Isue with help command line option.
• Addressed issue with Windows deleting temp directory.
• Issue with log retrieval. Last 2 rollovers of server logs, slow logs, and current+last 2 rollovers of gc logs now captured.
• Issue where X-pack related calls were not being run.

Support Diagnostics 6.3

6.3:

• Add network cache settings to diagnostics bundle #188
• Bring GC logs by default #190
• Remove config directory from diagnostics bundle #192

Support Diagnostic 6.2

Modified to fail gracefully and continue with archive if insufficient permissions to read log or config.
Fixed issue where numeric key in config file caused ClassCastException
Made iostat call iterative.

Support Diagnostic 6.1

• Updated JCommander to fix help indenting
• Adjustments to location of command to get stack trace
• Update console logging to show the text of each system command prior to running.

Support Diagnostic 6.0

• Support for Elasticsearch 6.0
• Updated documentation
• X-Pack queries only executed if it is installed.
• Two most recent archived logs included in addition to current log.
• --archivedLogs option now collects entire contents of log directory.
• Timed interval runs have the same timestamp for their names with run- appended.
• Added -type elastic-threads to collect only thread dumps for Elasticsearch
• Minor additions to API calls to add information.
• If multiple installations exist on the same host, config and logs will not be collected for those as well.
• --scrub option for logs temporarily removed.

Elasticsearch Support Diagnostic 5.1.3

• Settings displayed in flattened format.
• Added build instructions.
• shards.json displayed in bytes.
• nodeId added to cat_nodes
• Logstash option collects plugins
• Fixed issue with custom logstaash port
• Calls for machine learning
• Jstack call added to system calls
• Missing major version displays error message
• Call to /_cat/count removed. Already in indices_stats, and errors out if >1k shards.
• Added detailed to tasks call.
• Modified message for password prompt
• Added documentation for timed runs
• Removed minimum interval check for timed runs.
• Added flag to take only hot threads and jstack calls.

Elastic Support Diagnostics 5.12

Fixed issue with default config locations in package installations. Log settings for the defaults are moved to the settings/path tree but not configs so they were being missed causing both the config retrieve and the scrub to fail.

Elastic Support Diagnostics 5.11

Added new call for 5.x systems. Added include_defaults to cluster settings call to dump out all existing settings in force as well as ones set dynamically by users.