Supporting Atlassian Software

Editor's Notes

• #2: * Supporting Atlassian Software - brief how-to\n* No magic\n* Principles for managing any application - and more specifically a Java web application - apply\n* Reading 'The CheckList Manifesto' by Dr Atul Gawande - main message: no matter how expert you may be, well-designed check lists can improve outcomes in nearly all circumstances - examples of surgical teams, pilots.\n* Some of these points could be captured in your Support Checklist\n
• #3: \n
• #4: * Nature of Atlassian software - grows ORGANICALLY from small roots - small teams grow across organisation\n* Important to document: INSTALLATION, WHATS CONNECTED and HOW (in case JIRA admin heads off to the Bahamas)\n* CHANGE LOG - capture all changes - PATCHES, PLUGINS, INFRASTRUCTURE changes\n* Some clients - COMPLETE DOCUMENTATION - rebuild deployments from scratch - STEP BY STEP\n* UP TO DATE - INCORRECT worse than NONE\n* STORE WISELY - app containing info may be down\n
• #5: \n
• #6: * DEFINE YOUR POLICY TODAY\n* More software used - more important information is stored - MORE CRITICAL it is use to build in RESILIENCE to usage\n* Get Backup policy right from START\n* Consult with your IT dept - should fit with their normal procedures\n* NATIVE DB tools should be used in most case - XML may become corrupted and unable to restore - uses a lot of memory in its generation\n* Feature request: Confluence WILL NOT WARN you when BACKUP FAILS\n
• #7: \n
• #8: * AGGRESSIVE release policy - new features/bugs/improvements available ASAP (6 JIRA releases, 8 Confluence releases, 10 Bamboo releases)\n* Can be too QUICK for some (integration with other systems, support) - ADVISABLE to stay as up to date as your IT team will PERMIT\n* EOL policy - supports major versions for two years after the last minor iteration of that version was released (e.g. Confluence 2.2.x was supported for 2 years after Confluence 2.2.10 was released)\n* STAY AHEAD OF EOL WAVE - avail of improvements and security patches if nothing else\n* TEST - Checklist for what must work - JIRA (workflows, custom fields), Confluence (theme, permissions)\n* Plugin compatibility checklist - are all your key plugins compatible with latest version\n\n
• #9: \n
• #10: * BENEFIT - upgrades get the lastest security fixes and patches\n* SECURITY ADVISORIES - vulnerabilities or breaches found within their applications - documentation/mailing lists - managed via my.atlassian.com\n* LAST NIGHT - Advisory for JIRA with 2 high level XSS/XSRF issues and a moderate and low risk vulnerabilty on secure tokens and components\n
• #11: \n
• #12: * EXACT CLONE OF PRODUCTION - otherwise you lose a lot of the value\n* Never SURE if change/fix/plugin will perform same WAY on PRODUCTION\n* SANDBOX where you can test INFRASTRUCTURAL changes\n* New PLUGINS - verify FUNCTIONALITY / COMPATIBILITY / PERFORMANCE\n* STRESS / LOAD testing - LOAD REPLICATION TOOL\n* EXISTING COMMERCIAL lisense holders = DEVELOPER LICESNE - no need for full commercial license\n
• #13: \n
• #14: * PLUGINS - make ECOSYSTEM so INTERESTING AND VIBRANT\n* Applications not just ISSUE TRACKER or WIKIS - PLATFORMS to BUILD ON and INTEGRATE with other systems\n* Spiderman's Grand Dad - WITH GREAT POWER COMES GREAT RESPONSIBILITY\n* NO STEVE JOBS to curate what appears in your UPM\n* DETECTIVE WORK - docs / release/ compatibility / source code\n* TEST ENVIRONMENT - experiment with plugin\n
• #15: \n
• #16: * START GOING WRONG - the LOGS are your friends\n* INDICATE if a problem is BREWING or if it has already HAPPENED\n* PINPOINT where problem occurs - or at least good indication\n* GENERAL HOUSE-KEEPING - weed out over-zealously reported expcetions (macro wrong param)\n* VITAL for support support request - first response will be to ATTACH the logs\n* Eash through Confluence - SUPPORT UTILTIY TOOL\n
• #17: \n
• #18: * MORE MONITORING\n* Resource monitoring - MEMORY / CPU / DISK\n* Application monitoring - RESPONSE TIMES / LOAD / CONNECTIONS / IO / DATABASE TRENDS / USER ACCESS LOGGING\n* GATHER over PERIOD of TIME - VISUALISE DATA - see trends\n* TOOLS - Hyperic / Pingdom\n\n
• #19: \n
• #20: * WHEN APPLICATIONS GO BAD\n* The SCENARIO that should have NEVER happened in REAL LIFE - bubbles up to user in all its glory!\n* CAPTURE as much data as possible (LOGS, steps to REPRODUCE, SCREENSHOTS)\n* Logs are invaluable - user access logging, you should be able to see exactly what the user was doing\n* SERIOUS PLUGIN ISSUE - disable until diagnosed and fixed\n
• #21: \n
• #22: * ALL SYSTEMS hav LIMITED RESOURCES - MEMORY is one of them\n* 32-bit Windows system - 1.4GB-1.6GB - 64-bit system provides more\n* Sometimes INCREASING MEMORY can help - but NOT ALWAYS!\n* Can cause more problems - especially if there is a MEMORY LEAK!\n* GC Logging / HEAPDUMPS on OOME\n* MAKE ONE CHANGE at a TIME\n
• #23: \n
• #24: * REPORTS of SLOW RESPONSE times\n* PROFILING to see brekdown of where time is spend on EACH REQUEST\n* Thread dumps can also show exactly what the application is working on if slow or not responding\n* PROFILING TOOLS - YOURKIT\n
• #25: \n
• #26: * ALL A LITTLE OVERWHELMING - help is out there!\n* Atlassian offer an excellent support service\n* A lot of ONLINE DOCUMENTATION - KNOWLEDGE BASE\n* COMMUNITY SUPPORT - Forums - Altassian, Adaptavist - many partner sites\n* PARTNER PROGRAMS for SUPPORT \n
• #27: * ADAPTAVIST SUPPORT\n* Comprehensive diagnosis and advice service for all your JIRA, CONFLUENCE or CROWD questions/issues\n* Can also help out in EOL situations incase you have been submerged by that wave.\n
• #28: \n
• #29: \n