Releases · elastic/support-diagnostics

22 Jul 18:31

nemonster

Mainly adding some minor fixes and 5.0 specific handling.

Fixed formatting in some files and extensions.
Added 5.0 specific chain of calls including new ones for ingest/pipelines and cluster allocation explain.
- Cluster allocation explain only makes the default call that gets the data on why the first unassigned shard failed since each one is a separate call. If it turns out there are multiple reasons it's possible to go for a multi-index run and take a selection of 5 or so.
Refactored some of the query code to reduce duplication after adding the Logstash queries routine.

Assets 3

08 Jul 20:26

nemonster

Added this release:

Updated Logstash diagnostic API calls to 5.0-alpha-4 versions.
Support for client cert authentication via --keystore and --keystorePass arguments
Added calls to get user and role information from native realms.
Fix for Windows bat script to add enabledelayedexpansion.
jps added to local script execution.

Assets 3

16 May 19:37

nemonster

In this release:

Fixed issue where if a system call did not exist on the system would short circuit subsequent calls. Calls will continue now when any of them in the sequence fail.
All calls producing json results now have ?pretty&human.
fixed case where cat_fielddata was a non-json result but had .json extension.
call for _template added.
added option for LS 5.0 for REST diagnostic retrieval.
- --type logstash
- default port is 9600, no need to specify if using default.
- Currently emits errors after first retrieval for all subsequent calls. Not an issue since it will retry, but be aware that it's going to happen.

Assets 3

18 Apr 19:02

nemonster

Added option to create bzip rather than gzip for cases where more compression is needed.
Added diagnostic checks: node attrs, watcher stats, watcher stack traces
Empty yaml config files generate messages and don't necessarily error out.
Missing diag and chains yml generate errors with explicit messages
Invalid hostname now flagged.
Link to archive fixed in README.

Assets 3

30 Mar 20:45

nemonster

Number of changes/additions in this one. KB will be updated with expanded coverage of some features.

REST calls were refactored to use only the HTTP Client and remove the Spring REST Template. Done to optimize the process of writing the results to disk and some SSL handling. So much lower level work was getting done it was no longer an advantage to use the convenience wrapper.
Large cluster state should be be less of an issue for the default heap.
Errors such as 404's will be logged but the long stack traces are not displayed.
Verification of hostname in server certificate can now be bypassed.
Passwords and AWS access/secret keys now redacted from the elasticsearch.yml automatically. Anything having an identifier containing 'secret', 'access', or 'password" will have the value replaced with asterisks.
A log scrubbing facility is now bundled and activated with the --scrub argument. There is a scrub.yml in the diagnostics directory. Any value that you wish to replace in any of the log files or elasticsearch.yml should be entered in the key field, with the desired replacement string entered in the value side. It will walk the files line by line searching for instances of the keys to be scrubbed, replace them and then rewrite the file with "scrubbed-" prepended to the original name showing it was processed.
Added call for field data that breaks it down by node/index/field
Added call for watcher to dump stack traces of executing watches as discussed in 3/30 Bugpool.

Assets 3

01 Mar 14:19

nemonster

2.0.5

Added parameter -d --diagType with options standard(default) and remote. Using remote runs only the REST commands and does not run the system commands or retrieve logs and configs on the host it is being run from.
Changed logging so that all logs are generated in the working directory - in this case the installation directory. All logging is also directed to the console so that any errors are displayed in both the console and the logs.
Removed option for interactive choice of host to use. Host must be specified at runtime or diagnostic will display error and exit.
Modified fielddata check to break down usage by field and changed format to json.

Assets 3