Skip to content

Set of scripts for collection of diagnostic information from DSE/Cassandra clusters

License

Notifications You must be signed in to change notification settings

murilocmiranda/diagnostic-collection

 
 

Repository files navigation

This directory contains a set of scripts to generate a diagnostic tarball for DSE, DDAC & open source Cassandra installations, similar (and partially compatible) to diagnostic tarball generated by OpsCenter.

Generation of diagnostic tarball for DSE/Cassandra consists of 2 parts:

  1. Collection of data on individual nodes
  2. Merging all data into single file

These 2 steps could be separate because administrators may have different ways to access nodes in the cluster, and transferring the data. The collect_diag.sh performs data collection from cluster as one command, executed on one of the nodes of the cluster, or from Mac or Linux machine outside of the cluster.

collect_diag.sh

This script copies the collect_node_diag.sh file to all nodes, executing it in parallel, copying back collected data, and generate resulting tarball with DSE Insights or diagnostic data using generate_diag.sh.

Usage:

./collect_diag.sh -t <type> [options] [install_root_path]

This script accepts the same arguments as collect_node_diag.sh, with only -t as required parameter - all other are optional, but could be used to pass SSH, nodetool, cqlsh, dsetool options (for example, you need to pass these options if you have authentication enabled for cqlsh and/or JMX). For tarball installations it's required to pass the path to the top-level directory of the DSE/DDAC/Cassandra installation (pass -h to get a list of all options):

  • -B - AWS S3 bucket to upload the artifacts to
  • -c - specifies options to pass to cqlsh (user name, password, etc.)
  • -d - specifies the options for dsetool command (JMX user, password, etc.)
  • -f - specifies file with list of hosts where to execute collect_node_diag.sh (default - try to get list from nodetool status)
  • -e - key file for encryption of the generated tarball
  • -i - specifies that we need to collect insights data (DSE Metrics Collector and Metric Collector for Apache Cassandra)
  • -I - specifies directory that contains insights .gz files (default is /var/lib/cassandra/insights_data/insights)
  • -k - collect keystore and truststore information via Java's keytool
  • -K - AWS key for the S3 upload (Can be stored as the DS_AWS_KEY environment variable instead)
  • -m - specifies the collection mode: light, normal, extended (default: normal). See section below for information on what is collected:
    • light - collect only necessary information - logs, schema, nodetool, base system information
    • normal - as previous, plus extended system information (iostat, vmstat, ...)
    • extended - all Cassandra logs, longer vmstat/iostat, etc. It could be significantly slower, and will generate bigger archive
  • -n - specifies a list of general options to pass to nodetool (JMX user, password, etc.)
  • -o - specifies directory where to put where to put resulting file (default: automatically created)
  • -p - specifies the PID of DSE/DDAC/Cassandra process. Script tries to detect it from the output of ps, but this may not work reliably in case if you have several Cassandra processes on the same machine. This PID is used to get information about limits set for process, etc.
  • -P - top directory of COSS, DDAC or DSE installation (for tarball installs)
  • -r - remove collected files after generation of resulting tarball
  • -s - options to pass to SSH/SCP
  • -S - AWS secret for the S3 upload (Can be stored as the DS_AWS_SECRET environment variable instead)
  • -t (required) - specifies the type of installation: dse, ddac, coss;
  • -T - ticket/ref id for the S3 upload and encrypted tarball naming (required for S3 uploads)
  • -u - specifies timeout for SSH in seconds (default: 600). You may need to increase it if you're using extended collection mode
  • -v - enables more verbose output by all scripts
  • -z - don't execute commands that require sudo

Please note that user name should be passed as -o User=... in -s option, as scp and ssh are using different ways to pass user name.

Example of usage with list of hosts passed explicitly (file mhosts):

./collect_diag.sh -t dse -f mhosts -r -s \
  "-i ~/.ssh/private_key -o StrictHostKeyChecking=no -o User=automaton"

or for DDAC:

./collect_diag.sh -t ddac -f mhosts -r -s \
  "-i ~/.ssh/private_key -o StrictHostKeyChecking=no -o User=automaton" \
  -P /usr/local/lib/cassandra

if it's running on the machine with DSE/DDAC/C*, then it could be as simple as, as it will use nodetool status to obtain a list of nodes:

./collect_diag.sh -t ddac -r -P /usr/local/lib/cassandra

In order to encrypt the generated tarball and upload it to S3, specify the necessary arguments:

./collect_diag.sh -t coss -K <aws key> -S <aws secret> -T <ticket id> -e /path/to/enc_secret.key -B <s3 bucket name> -r

What is collected

The collect_node_diag.sh script collects following information:

  • all DSE/DDAC/Cassandra configuration files - cassandra.yaml, dse.yaml, /etc/defaults/dse
  • all current log files in the light & normal collection modes, all log files, included rotated, in the extended mode
  • data from nodetool sub-commands, like, tablestats, tpstats, etc.
  • data from dsetool commands - at least status & ring
  • (in extended mode) executes for DSE the nodetool sjk mxdump to get all current values in JMX;
  • database schema
  • schema and configuration for DSE Search cores
  • system information to help identify the problems caused by incorrect system settings (you may need to install some tools, like, iostat, vmstat, etc.):
    • information about CPUs, block devices, disks, memory, etc. (primarily from /proc filesystem)
    • information about operating system (name, version, etc.)
    • limits for user that runs DSE/DDAC/Cassandra
    • output of sysctl -a, dmesg, etc.
    • IO and VM statistics via iostat and vmstat (only in normal & extended modes)

Important: The generate_diag.sh script removes all sensitive information, such as, passwords from configuration files.

Collecting diagnostic on individual nodes

Collection of the data on individual nodes is performed by collect_node_diag.sh script that executes different commands to collect data described in the previous section. Script should be executed on every node of cluster, and if it's a tarball installation, you need to pass one required parameter - full path to the root directory of DSE/DDAC/Cassandra installation. For package installation, location of the files will be detected automatically, without specification of the root directory. There are also optional parameters, that could be provided if, for example, you have authentication enabled for Cassandra or JMX, changed JMX port, etc. (pass -h to get list of options):

  • -c - specifies options to pass to cqlsh (user name, password, etc.)
  • -d - specifies the options for dsetool command (JMX user, password, etc.)
  • -f - specifies the name where it should put the collected results (could be useful for some automation)
  • -i - specifies that we need to collect data for insights data (DSE Metrics Collector and Metric Collector for Apache Cassandra)
  • -I - specifies directory that contains insights .gz files (default is /var/lib/cassandra/insights_data/insights)
  • -k - collect keystore and truststore information via Java's keytool
  • -m - specifies the collection mode: light, normal, extended (default: normal). See section below for information on what is collected:
  • light - collect only necessary information - logs, schema, nodetool, base system information
  • normal - as previous, plus extended system information (iostat, vmstat, ...)
  • extended - all Cassandra logs, longer vmstat/iostat, etc. It could be significantly slower, and will generate bigger archive
  • -n - specifies a list of general options to pass to nodetool (JMX user, password, etc.)
  • -o - specifies directory where to put where to put resulting file (default: /var/tmp/)
  • -p - specifies the PID of DSE/DDAC/Cassandra process. Script tries to detect it from the output of ps, but this may not work reliably in case if you have several Cassandra processes on the same machine. This PID is used to get information about limits set for process, etc.
  • -t - specifies the type of installation: dse, ddac, coss (default: dse)
  • -v - enables more verbose output by all scripts
  • -z - don't execute commands that require sudo
  • -P - path - top directory of COSS, DDAC or DSE installation (for tarball installs)"
  • -e - timeout - e.g. "-e 600" allow for a longer timeout on operations"

After successful execution, script generates file with name /var/tmp/dse-diag-<IP_Address>.tar.gz, like, /var/tmp/dse-diag-10.200.179.237.tar.gz, or into specified by option -f (or if the -w flag was used, the file name prefix will be ddac-diag-...).

Merging all diagnostics into single tarball

After the collect_node_diag.sh script was executed on every machine of the cluster, generated files should be collected into single directory on one machine to be merged using the generate_diag.sh script. This script accepts single parameter - path to directory with collected files (it could be either relative, or absolute). There are also optional parameters:

  • -f - specifies path to file where data should be put
  • -i - specifies that we need to merge Insights data
  • -o - specifies directory where to put where to put resulting file (default: /var/tmp/)
  • -p - specifies the custom pattern for file names if collect_node_diag.sh was called with -f parameter - otherwise this script may not find the data
  • -r - remove individual diagnostic files after processing
  • -t - specifies the type of installation: dse, ddac, coss (default: dse)

This script performs following:

  • Creates a temporary directory in the /var/tmp or specified by -o flag;
  • Unpacks each of collected files;
  • Removes sensitive information, such as, passwords from configuration files;
  • Packs everything together into single file that has name <cluster_name>-diagnostics.tar.gz, for example, dsetest-diagnostics.tar.gz, or into specified by -f option.

This file could be then sent for analysis to DataStax support, or analyzed by tools, like, sperf.

Encrypting and uploading the diagnostic tarball to S3

If the instance running collect_diag.sh doesn't have access to S3, encrypting and uploading the tarball can be done afterwards using the encrypt_and_upload.sh script (pass -h to get a list of all options):

  • -f - (required) diagnostic file to encrypt and upload to S3
  • -B - (required) AWS S3 bucket to upload the artifacts to
  • -T - (required) ticket/ref id for the S3 upload and encrypted tarball naming
  • -e - key file for encryption of the generated tarball
  • -K - AWS key for the S3 upload (Can be stored as the DS_AWS_KEY environment variable instead)
  • -S - AWS key for the S3 upload (Can be stored as the DS_AWS_SECRET environment variable instead)

If the encryption key is not provided, it will be generated by the script. Here's how to encrypt and upload a tarball to an AWS S3 bucket:

./encrypt_and_upload.sh -f /path/to/tarball.tar.gz -K <aws key> -S <aws secret> -T <ticket/ref id> -e /path/to/encryption_secret.key -B <s3 bucket name>

Contact

For questions, to report an issue, to request a feature/enhancement, etc., please use GitHub Issue or visit our Community Forums: https://community.datastax.com

License

Copyright DataStax, Inc.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.


DataStax is a registered trademark of DataStax, Inc. and its subsidiaries in the United States and/or other countries.

Apache Cassandra, Apache, Tomcat, Lucene, Solr, Hadoop, Spark, TinkerPop, and Cassandra are trademarks of the Apache Software Foundation or its subsidiaries in Canada, the United States and/or other countries.

About

Set of scripts for collection of diagnostic information from DSE/Cassandra clusters

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Shell 86.3%
  • Python 10.8%
  • Gherkin 2.9%