cscott/ask-annie
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add chapters for session 754694580 (Haiku)

Browse Source
This commit is contained in:
Conan Scott
2026-03-24 09:58:52 +00:00
parent 7dc7998475
commit eae563ad47
1 changed files with 18 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
chapters/754694580.json Normal file
View File

@@ -0,0 +1,18 @@
[
{"timestamp": "0:26", "title": "Directory Structure Service Enabled Configuration", "summary": "The Directory Structure Service Enabled parameter caches subscription folder information at login instead of rechecking all folders every time. First login takes same time as before; subsequent logins are faster. Enable for accounts that log in frequently with many subscriptions."},
{"timestamp": "3:38", "title": "HTTP vs SSH Folder Checking Behavior", "summary": "SSH only checks the current folder where files are placed. HTTP loads the entire directory structure on login. Directory Structure Service caching is more impactful for HTTP deployments."},
{"timestamp": "5:46", "title": "Scheduled SSH Transfer Configuration Recommendations", "summary": "Enable Directory Structure Service for accounts with many subscriptions regardless of protocol, as it applies to both login and pull operations. Check folder existence during pulls affects performance."},
{"timestamp": "9:47", "title": "Cluster Failover Issue After OS Patching", "summary": "Secondary server self-declares as primary during cluster startup, likely caused by networking delay after OS patch. Solution: shut down both servers, start primary, wait for 'become primary' log entry, then start secondary."},
{"timestamp": "14:27", "title": "Cluster Patching Order: Primary vs Secondary", "summary": "Patch order doesn't matter functionally. Recommendation: shut down entire cluster, patch primary first while it's the only server, then patch secondary. Avoids flip-flop where secondary becomes primary and transfers tracked on different table."},
{"timestamp": "18:39", "title": "Windows Automatic Service Startup Ordering", "summary": "Windows automatic services start too fast without proper ordering. Database must start before transaction manager. Recommendation: disable automatic startup and create separate service that runs 'start all' late in boot sequence to allow networking services to stabilize."},
{"timestamp": "21:32", "title": "Stop All Command Java Process Cleanup", "summary": "In 5.5 releases, 'stop all' stops all processes including Java instances from ST folders. If Java process remains after stop all, startup had a problem and process disconnected. In old releases, schedulers required manual killing; this is fixed in 5.5."},
{"timestamp": "24:03", "title": "VIP Failover Configuration and Port Monitoring Loss", "summary": "In 5.5, TM communication direction reversed; listener disappeared, removing easy VIP port-monitoring capability that existed in 5.0.1. Recommendation: run scheduler checking cluster status and controlling unused protocol port for failover detection instead."},
{"timestamp": "28:06", "title": "Active-Active Cluster Architecture vs VIP Limitations", "summary": "ST in active-active mode has both nodes accepting inbound and processing outbound. Primary handles 60-80% of jobs; secondary handles 30-40%. VIP in failover mode limits to one server receiving inbound traffic, overloading primary. Load balancer with weighted distribution recommended."},
{"timestamp": "31:46", "title": "Secondary Server as Primary Inbound Destination", "summary": "In active-active cluster with VIP, consider routing inbound to secondary (slower server) and outbound to primary. Prevents overloading primary with both server-side jobs and client inbound connections. Requires marking secondary as primary in VIP terms for failover."},
{"timestamp": "35:04", "title": "Cluster Failover Detection Workaround", "summary": "Create scheduler service checking who is primary every 2-5 minutes. If not primary, shut down unused protocol port (e.g., PECID, AS2). VIP detects port down and switches traffic. Provides semi-automatic failover with minutes of downtime instead of requiring manual intervention."},
{"timestamp": "40:04", "title": "Primary Cluster Node Detection Methods", "summary": "Multiple methods to detect primary status: check 'become primary' log entry, monitor temp folder sentinel file, use API call, or attempt SSH/HTTP login. Log entry provides most reliable detection mechanism for scripts."},
{"timestamp": "43:20", "title": "FTP ASCII Mode Line Terminator Transcoding", "summary": "ASCII mode tells receiving server to strip line terminators and apply its own format. Windows CRLF converts to Unix LF; opposite occurs Unix to Windows. Receiving server handles conversion, not sender. Binary mode preserves file as-is."},
{"timestamp": "46:06", "title": "AutoDetect vs ASCII Mode File Transfer", "summary": "AutoDetect uses MIME types from OS level to determine transfer mode. Text files treated as ASCII; binary files transferred as binary. ASCII mode behavior should match AutoDetect for text files. If inconsistent, open support ticket."},
{"timestamp": "52:22", "title": "Tracking Table File Mode Verification", "summary": "Tracking table shows actual transfer mode used (ASCII or binary). Verify correct mode was applied by checking outbound tracking entry. If mismatch with expected mode, indicates a bug."},
{"timestamp": "54:02", "title": "ASCII Mode Configuration Behavior", "summary": "Setting transfer site to ASCII mode forces all files through that site to use ASCII connection regardless of file type. Does not modify file contents; sets protocol connection mode informing receiving server how to handle line terminators. Safe to enable for text-only sites."}
]