FrequentlyAsked Question / Exact Problem
- “Journey shows Failed status even though we ran it for 7.5 lakh customers.”
- “Campaign delivery count shows 53K but Karix says 4 lakh delivered.”
- “Delivery report for Chumbak shows read/delivered/clicked data missing.”
- “FirstConnectt SMS delivery count is doubled — targeted 8 but delivered shows 16.”
- “Pepe delivery report showing incorrect numbers.”
Affected Tenant / Context
Pepe Jeans (Karix/FirstConnectt SMS), Chumbak (Qikchat WhatsApp), Sathya (Robeeta SMS),
Mufti, Otto, General across tenants using external message providers.
Root Cause
Multiple causes identified across incidents:
- ClickHouse connection pool exhausted — For very large journeys, CH connections were
not being returned to the pool when errors occurred. This caused the journey to fail midrun. Fix: added proper catch block to return connections on error. - IoT Web API downtime — Delivery callback webhooks from Karix to CASA’s IoT Web API
were being dropped because the API was down. 44K out of 97K delivery responses were lost. Creating a new service replaced the old one and prevented recurrence. - Qikchat delivery failure — template parameter mismatch — Chumbak’s WhatsApp
campaign failed because the template was created with a video attachment in Qikchat but CASA was passing an image in the API call. Qikchat rejected the message and didn’t send delivery callbacks. - FirstConnectt duplicate delivery callbacks — For SMS messages over 160 characters,
FirstConnectt splits them into 2 parts and sends a delivery callback for each part. CASA counted both as separate deliveries, inflating the delivered count. - NullPointerException in Infersales — A code-level NPE in the Infersales component
prevented journey “Conversion” marking for some tenants (Mufti, Chumbak). Dineshr
fixed and reprocessed the affected queue.
Common Debugging Steps (For CS Team)
- Check JourneyAnalytics in the dashboard. Note: Journey Builder count ≠ Campaign
Analytics count because JourneyAnalytics uses unique passed_rule_id while
Campaign Analytics uses unique contact_mobile . Some difference is expected. - For WhatsApp campaigns: log into the Qikchat portal and check delivery status there. If messages show as “Failed” in Qikchat, the issue is a template mismatch — CS can verify this directly without dev.
- For SMS campaigns via FirstConnectt: if delivered count is double the targeted count, check if the SMS template is over 160 characters.
- If a journey shows “Failed” status (not just low delivery): escalate immediately with Journey URL and Journey ID.
- For Chumbak/Qikchat: ask the provider directly for delivery report CSV if the analytics are missing, then share with CASA dev to upload manually.
Fix Made & Developer Conclusion
- Pepe journey failed: ClickHouse connection pool fix deployed. Journey re-triggered next
morning. For large journeys (7.5 lakh+), minimum wait time between retriggers set to 2
hours. - Pepe delivery count mismatch (53K vs 4L): IoT Web API was down during the campaign.
New service created. Historical data requested from Karix via SFTP to manually upload
missing delivery statuses. - Chumbak Qikchat: Template parameter mismatch confirmed. CS team now trained to
verify template in Qikchat portal before escalating. - FirstConnectt double-count: Confirmed as provider behaviour (160-char splitting).Provider must handle deduplication on their end. CASA team informed. Taruun Prasath GS fixed the delivery status mapping issue (field mismatch in webhook parsing).
- NullPointerException/Conversions: Fix deployed by Dineshr. Affected queue
reprocessed. Conversion data recovered.
Exact Dev to Reach Out To
Dineshr — Journey execution, queue, Qikchat/Karix integration
Taruun Prasath G S — Delivery webhook, notification service, FirstConnectt
Dharshini — Journey analytics, NullPointerException in Infersales, delivery report
Vishnuj — IoT Web API, Karix SFTP integration