SLDG Shield · Sep 29 → Oct 6

Renê Cardozo

Jessica Sagen

7issues, 2 customer-facing 72 → 18production pings vs last week, ~15 pages 11open follow-ups, 4 without a ticket

Week overview

Navan launch: supplier availability and confirm errors

Navan went live Sep 29. Omni confirm errors fell for every supplier we fixed; WebBeds and Nuitee are still above target.

Effect

Navan bookings199 → 471Oct 5 → Oct 6

Navan supplier availability

onpartial / restrictedoff


Amadeus	19.7%	2.6%	Session opened at ConfirmRate
Hotelbeds	25.6%	0.9%	RECHECK rates dropped
WebBeds	22.8%	8.8%	Direct block; still above target
Nuitee	20.1%	31.9% n=113	Back after resort-fee hold
Sabre	3.7%	3.5%	Confirm stable; FBO spiked on return
AVH	22.3%	off	Off for Omni until Pricing fixes the tax gap

Details and evidence

Before is 18h and after is 23h; rates are summed errors ÷ summed responses. Target ≤3%; Navan expects ~3.4% failures. confirmRate_error ÷ confirmRate_response, service-group:partner (includes other Omni customers). Navan bookings by Oct 6, 08:00: 199 on Oct 5 → 471 on Oct 6, Navan resort-fee comms done.

Supplier on/off timeline from the #omni-launch-war-room updates

Alerting standardization and the move to incident.io

Problem

Monitors paged on raw counts in every environment, through two paging tools.

Action

Search alerts moved to error rates with volume gates, and paging moved to incident.io only SLDG-3109.

Effect

Production pings / week72 → 18dev + staging 23 → 20Paging tools2 → 1incident.ioProduction pings still non-actionable7 of 18details

Production pings per Shield week

Tuesday to Tuesday handoffs

#sapi-lodging-alerts ↗

Follow-ups (3)

engine-search#1459SLDG-3152 Container restarts monitor standard · merged Oct 6, after the handoff[SLDG-3108] Omni FA dependency monitor on an error rate · PR closed unmerged; threshold only raised in Datadog, Terraform still says 100Keep dev smoke tests out of the no-offer P1 monitor · no ticket · fired 9× this week

Details and evidence

SLDG monitors paged on raw counts across dev, stg and prd, and through two paging tools (OpsGenie and incident.io). Production pinged Shield 31, 64, 147, 143, 72 times in the weeks before this one.

Search monitors unified on service_group with error rates and volume gates ([SLDG-3105] Unify lodging alerts for the Navan ramp, [SLDG-3029] Alert on the no-offer rate). Booking INTERNAL counts removed ([SLDG-3106] Remove booking INTERNAL count alerts). Paging via incident.io only ([SLDG-3109] incident.io-only paging), with the schedule managing @sapi-lodging-shield. Integrations monitors handed to SPI ([SLDG-3149] Integrations monitors to SPI).

How pings are counted: Bot posts whose title starts with Triggered, Re-Triggered, Warn, No data, Alert or P1–P3 (not Recovered, not [SPI), duplicate Datadog posts within 2 minutes counted once; dev/staging split by the SLDG-dev/SLDG-stg, phase or instance markers.

CCA & Sertifi

Sertifi was on for far more hotels than use it; see Sertifi auto-enable.

Effect

Sertifi-enabled hotels11,3337,183 at V2 launch; rest auto-enabledForms processed by hotels25.7%31.7% opened · Jul 28 → Sep 30Sertifi auto-enableon → offunused properties disabled + ADMIN-locked

Details and evidence

Details in Sertifi auto-enable. Other CCA asks this week: [BUGS-132] send error, Primecap Sertifi transactions (DM), CCA signals contract, Canary CVC, [DPLAT-2557] CCA events in Snowflake.

Supplier logs follow one naming scheme across repos

Problem

Each supplier repo named its payload logs differently, so failures were hard to find.

Action

One naming scheme, <supplier>.<engineOp>.<step>.<supplierCall>.<request|response|error>, now used across he-relay, the shared libraries and the Amadeus REST, DerbySoft and Nuitee repos.

Effect

Supplier code bases on the scheme1 (partial) → 4 + shared libsFailures in 2xx bodies.response → .errorCard data in payload logsmasked

Details and evidence

Supplier payload logs were named differently in each repo. he-relay mostly followed a scheme but had hops without a step, invented error steps, and 2xx failures logged only as .response (audit in [SLDG-3062] Hop business errors under the hop step). The Amadeus REST, DerbySoft and Nuitee repos didn't follow it.

One scheme, <supplier>.<engineOp>.<step>.<supplierCall>.<request|response|error>, across he-relay, the shared lodging libraries and the Amadeus REST, DerbySoft and Nuitee repos:
• he-relay: [SLDG-3062] Hop business errors under the hop step · [SLDG-3066] WebBeds and Priceline errors under the calling hop
• Shared libraries: [SLDG-3026] Shared gRPC payload log names · [SLDG-3059] captureSupplierCall through the shared resolver · [SLDG-3056] Envelope log names like he-relay
• Supplier repos: [SLDG-3005] Amadeus REST payload log names · [SLDG-3006] DerbySoft payload log names · [SLDG-3007] Nuitee payload log names
• Card masking: [SPI-346] Mask card data in payload logs

Tickets: [SLDG-3027] Modification type in hop log names · [SLDG-3061] Shared supplier payload resolver. Naming rule: he-relay docs/agents/TypeScript.md.

Issues & resolutions

Navan confirm failures: supplier sessions expired before confirm

Customer-facingSep 29 → Oct 1SLDGSPIP1Navan report

Problem

Amadeus and AVH sessions opened at FA expired after 10 minutes, so late confirms (~14% of bookings) all failed.

Action

Open the session at ConfirmRate SPI-420, behind a flag, live Oct 1, 11:47.

Effect

Amadeus Omni confirm errors19.7% → 2.6%Confirms after 10 min failing100% → 0%Amadeus confirm latency+1.5s

Omni Amadeus confirmRate error rate (%, 6h buckets)

errors ÷ responses · service-group:partner · prd

Datadog ↗

Follow-ups (2)

he-relay#2296SLDG-2689 AVH child ages on Hotel_Sell · merged Oct 6, after the handoffAVH tax fallback at confirm · waiting on Pricing's estimate

Details and evidence

Amadeus/AVH sessions opened at FA expired after 10 min. Navan confirmed late on ~14% of bookings, all of which failed, and Navan reported 17% errors.

Open the session at ConfirmRate behind a flag ([SPI-420] Open Amadeus sessions at ConfirmRate, [SPI-421] Open AVH sessions at ConfirmRate), live Oct 1, 11:47.

PRs: [SPI-418] token data · [SPI-419] fresh session pricing · [SPI-416] Amadeus session hygiene / [SPI-417] session hygiene · flag [SPI-414] Session-at-ConfirmRate flag in dev, stg, sandbox, [SPI-414] Session-at-ConfirmRate flag in prd · release v2.69.0
Tickets: [SLDG-3122] Supplier sessions at ConfirmRate · [SPI-421] AVH sessions at ConfirmRate
Threads: 4-tracks plan · launch · result
Members vs Omni confirm success before the fix: Amadeus 99.2% vs 91.4%.

Navan resort-fee incident: fees charged at booking or missing

Customer-facing · Navan threatened to pause EngineOct 2 → Oct 5SPIOmniSupplyClient OpsP1cross-teamstatus update

Problem

Sabre resort fees (code 12) were stored as prepaid and charged at booking, and Nuitee left fees off some rates. 939 bookings were at risk.

Action

Navan ran on Amadeus only while bookings were fixed; Sabre fees are now postpaid SPI-480, live Oct 3, 10:40.

Effect

Sabre fees classified postpaid0 → 17k–124k per 6hNavan suppliers1 → 5Oct 3–5 → Oct 5Navan bookings by 08:00199 → 471Oct 5 → Oct 6Nuitee Omni confirm errors31.9%n=113

Omni confirmRate volume by supplier (6h)

Shows suppliers switched off and back on for Navan

Datadog ↗

Follow-ups (3)

[SPI-483] Navan Sabre code-12 audit: 102 of 854 bookings, $10,016[SPI-474] Nuitee: backfill missing resort fees from the tax-schema endpoint · promised for Oct 5, no PR yetNavan resort-fee comms

Details and evidence

Sabre code-12 resort fees were stored as prepaid, so we charged them at booking; Amadeus treats the same code as postpaid. Nuitee also omitted fees on some rates. 939 bookings at risk, $10k–$70k.

Navan ran on Amadeus only while bookings were remediated. Sabre fees are now postpaid ([SPI-480] Sabre resort fees (code 12) postpaid, live Oct 3, 10:40). Nuitee came back once it agreed to return its own fee breakdown.

Ticket: [SPI-480] Sabre resort fees postpaid · prod deploy
Threads: Navan question · Nuitee non-sellable · re-enable plan · Nuitee 100% re-enabled (Oct 6, 01:23) · WebBeds back (Oct 5, 12:28)
Slack has no explicit "Sabre/Nuitee re-enabled for Navan" message; the Omni traffic in the chart is the evidence. The Omni group also includes other Omni customers.

Hotelbeds: RECHECK rates confirmed without checkRate

Omni degradation · pulled from Navan Sep 29Sep 29 → Oct 5SLDG SearchSPIcross-teamthread

Problem

Omni traffic skipped checkRate, so stale RECHECK prices failed at confirm.

Action

Omni traffic only gets BOOKABLE rates SLDG-3134, live Oct 2, 12:38.

Effect

Omni confirm failures25.6% → 0.9%Hotelbeds rates offered to Navan−81%only 19% are BOOKABLEMembers with checkRate (reference)3–9%

Hotelbeds confirm failure rate: Omni (no checkRate) vs Members (checkRate)

6h buckets · buckets with <20 confirms hidden

Datadog ↗

Follow-ups (1)

Re-shop at confirm (Hotelbeds/WebBeds) · stalled: WebBeds attempt rolled back Oct 2, Hotelbeds attempt closed

Details and evidence

Hotelbeds returns BOOKABLE rates (price guaranteed) and RECHECK rates (price must be re-checked with checkRate). Members calls checkRate; Omni skipped it because Hotelbeds caps checkRate and won't raise the limit. Stale RECHECK prices then failed at confirm.

Search sends an Amplitude-controlled header and the gateway drops RECHECK rates for Omni traffic ([SLDG-3134] Search flag to drop Hotelbeds RECHECK rates, [SPI-469] Gateway drops Hotelbeds RECHECK rates), live Oct 2, 12:38.

Tickets: [SLDG-3134] Drop Hotelbeds RECHECK rates · [SPI-469] Gateway drops Hotelbeds RECHECK rates
Threads: RECHECK rates on Omni · filter rollout · always-checkRate test dropped (rate limits break unrelated confirms)

WebBeds: room blocking failed at confirm

Omni degradation · pulled from Navan Sep 29Sep 29 → Oct 5SPI

Problem

Room blocks failed at confirm when WebBeds changed or renumbered the rate.

Action

Block with the offer-token allocation SPI-409, live Oct 2, 11:30; back for Navan Oct 5, 12:28 with a win-rate penalty.

Effect

Omni confirm errors22.8% → 8.8%Members confirm errors11.4% → 4.0%n=281 → 878

WebBeds confirm error rate: Omni vs Members

6h buckets · buckets with <20 confirms hidden

Datadog ↗

Follow-ups (1)

Re-shop at confirm (Hotelbeds/WebBeds) · stalled: WebBeds attempt rolled back Oct 2, Hotelbeds attempt closed

Details and evidence

WebBeds confirm means blocking the room. Blocks failed with 120003 No rooms found or 151–153 (rate or policy changed), and we couldn't match the blocked rate when WebBeds renumbered it ([SPI-401] Match the WebBeds blocked rate when runno changes).

Block with the allocation carried in the offer token ([SPI-409] Block WebBeds with the offer-token allocation, live Oct 2, 11:30). A confirm re-shop was tried and removed ([SLDG-3126] WebBeds confirmRate re-shop → [SLDG-3126] Remove the WebBeds re-shop). Back for Navan Oct 5, 12:28 with a win-rate penalty.

Tickets: [SPI-409] WebBeds direct blocking · [SPI-401] (runno renumbering, [SPI-401] Match the WebBeds blocked rate when runno changes) · [SLDG-3126] (re-shop)
Threads: re-shop design · bed-bank update (91.3% on Members)

Omni search timeouts and the Amadeus blip

Omni degradationSep 29 · Oct 2SLDG SearchSPI

Problem

Omni FBO hit the 6s supplier timeout and fired 7 alerts on Sep 29.

Action

Supplier timeout raised to 10s for Members and Omni.

Effect

Omni FBO integrations errors10–17% → 1–3%Hourly spikes after the fix15–19%Sep 30, Oct 1, Oct 2, Oct 5Oct 2 Amadeus blip839 errors / 5 min~10 min, self-recovered

Search: Omni findBestOffers requests hitting an integrations error (%, hourly)

integrations_client_find_best_offers_error ÷ request_count · service_group:partner · prd

Datadog ↗

Details and evidence

Omni FBO hit the 6s supplier timeout: Amadeus Omni p90 was 6.5s on 4x the Members volume, plus Sabre 429s. That fired 7 alerts on Sep 29.

Raised the supplier timeout from 6s to 10s for Members and Omni.

Mute / baseline thread · Oct 2 blip thread · monitors 238643003, 238642999

Sabre re-enable: Omni findBestOffers error spike

Omni degradation · self-recoveredOct 5, 06:00 → 16:00SPI

Problem

Sabre FBO errors spiked when its traffic returned, pushing Omni FBO errors to 15.6%.

Action

None needed; it recovered on its own.

Effect

Sabre Omni FBO errors~90/h → 923/hpeak 09:00Recovery<100/h from 11:00 · ~0–10/h from 16:00

Omni findBestOffers errors by supplier, Oct 5–6 (per hour)

Gateway findBestOffers_error.count · service-group:partner

Datadog ↗

Follow-ups (1)

he-relay#2293SPI-485 Sabre 10% commission default · approved, not merged

Details and evidence

When Sabre traffic returned, Sabre FBO errors spiked and pushed Search Omni FBO errors to 15.6%. SPI's Sabre alerts fired at 07:17 and 07:25.

None needed; it recovered on its own. A related Sabre commission default ([SPI-485] Default Sabre commission to 10%) is still waiting on a decision.

Sertifi auto-enabled on ~11k hotels: unused properties disabled and locked

Cost / operationsOct 1 → Oct 3SLDG BookingClient OpsProduct

Problem

Sertifi was auto-enabled on 11,333 hotels, but hotels processed only 25.7% of forms.

Action

Unused properties were disabled and locked off; auto-enable is now behind a per-provider flag SLDG-3147.

Effect

Sertifi auto-enableon → offUnused Sertifi propertiesdisabled + lockedcount pendingBaseline payment CI on Sertifi bookings2.31%Hyatt 3.14% · Marriott 2.94%

Follow-ups (2)

[DPLAT-2557] CCA delivery events in Snowflake (backfill) · accepted by Data Oct 1Canary: send the CVC with the VCC on CCA creation · no ticket

Details and evidence

CCA provider write-through enabled Sertifi on every property matched in Sertifi's portal, so 11,333 hotels were on Sertifi, 7,183 of them bulk-loaded at V2 launch. Hotels processed only 25.7% of forms, and Product flagged the cost ahead of the Sertifi renewal.

Properties that never opened or processed a form were disabled and ADMIN-locked off, and the config was synced to Salesforce. Auto-enable is now behind a per-provider flag ([SLDG-3147] Gate provider auto-enable per provider, [SLDG-3147] Sertifi auto-enable flag): off for Sertifi, on for Canary.

Data: [BA-1123] Sertifi delivery and CI data · Hex dashboard · Sertifi export (DM) (Hyatt opens most forms but rarely marks them processed)
Why it grew: sync maps every Sertifi property ID · count validation
Design: [PTD] CCA provider status ingestion · [SLDG-2373] Contract DTD: CCA delivery lifecycle events

Minor items (4) · fixed, handed off, or untriaged


Sandbox CCA reseed dead letters (Booking)	4 dead letters → fixed: blank stored totals treated as missing	[SLDG-3130] Blank stored amounts on the CCA path · alert
Hotelbeds Omni getBooking 53% failures	Navan sent invalid booking references; handed to SPI along with the monitor	alert · [SLDG-3149] Integrations monitors to SPI
Stg search pod CrashLoopBackOff	Single pod; self-recovered after ~36h, no root cause found	alert · [SLDG-3152] Container restarts monitor standard
Cache-warmer Key Sync "no data" (sabre-platform-v5)	Untriaged; monitor showed No Data on Oct 5	alert · monitor

Needs attention

Item
SLDG follow-ups7
Hotelbeds reconciliation after a Nexus cancel unassigned	SLDG	SLDG-3120
Omni FA dependency monitor on an error rate PR closed unmerged; threshold only raised in Datadog, Terraform still says 100	SLDG	SLDG-3108
Re-shop at confirm (Hotelbeds/WebBeds) stalled: WebBeds attempt rolled back Oct 2, Hotelbeds attempt closed	SLDG · SPI	link
FX backfill for existing Omni bookings in Salesforce asked twice by Finance; SLDG-3154 only covers future changes	SLDG	link
Keep dev smoke tests out of the no-offer P1 monitor fired 9× this week	SLDG	link
Reply to the RNPL refundability question unanswered since Oct 1; reply, no ticket needed	SLDG	link
Canary: send the CVC with the VCC on CCA creation	SLDG Booking	link
Waiting on other teams6
Sabre 10% commission default approved, not merged	SPI	he-relay#2293SPI-485
Nuitee: backfill missing resort fees from the tax-schema endpoint promised for Oct 5, no PR yet	SPI	SPI-474
CCA delivery events in Snowflake (backfill) accepted by Data Oct 1	Data Platform · SLDG	DPLAT-2557
AVH tax fallback at confirm waiting on Pricing's estimate	Pricing	link
“Air Conditioning” mapped to Omit	Content	waiting
Nomad async cancels with no SAPI visibility	Omni	waiting

Shield load

Pings in #sapi-lodging-alerts72 → 18vs last weekincident.io pages~15estimated from monitor routing: FA dep 7 · FBO dep 7 · latency 1Non-actionable7 of 18Real issues behind the pings10+1 untriaged

Production pings per day

One bar per UTC day, Sep 29 → Oct 6 handoff

#sapi-lodging-alerts ↗


Omni FA dependency errors	Chronic DerbySoft–Wyndham baseline on a raw count	6
Booking latency >1s	One slow call (Nexus CCA job)	1
Total		7

Smoke-test traffic: ~19k logs · 0 errors · 0 alerts · find_availability no-offer rate: 1.45%

Requests by team · 17 requests: 5 open · 1 unanswered · 2 handed off · 9 done

Ask and outcome
SPI1
AVH bookings fail on child occupancy Convert children to adults for AVH · [SLDG-2689] AVH child occupancy at Book, PR today	SPI	done
Omni3
`booking_modified` refund fields swapped Fixed with new refund objects ([SLDG-3116] Consistent customer and supplier refunds on events)	Omni	done
Refunded Nomad booking, Amadeus cancel timed out Nomad cancels async with no SAPI visibility · open with Omni	Omni	open
Does a Hotelbeds checkRate rate limit filter the offer? Yes, and it breaks unrelated confirms → test dropped	Omni	done
Omni / Business stats1
FX rates missing in Salesforce (DM) (INR read as USD) FX on booking events + Rocket mapping ([SLDG-3136] et al.); backfill ask open	Omni / Business stats	open
Members Checkout1
Nuitee create-cart errors Nuitee turned off; became the resort-fee incident	Members Checkout	done
Members Search2
FBO `otherAmenities` always empty Content mapping gap ("Air Conditioning" → Omit) · Content	Members Search	handed off
RNPL refundability differs PDP vs checkout No reply	Members Search	unanswered
Client Ops / Product1
Sertifi auto-enabled on ~11k hotels Unused properties disabled; auto-sync behind a flag ([SLDG-3147] Sertifi auto-enable flag)	Client Ops / Product	done
Client Ops1
[BUGS-132] CCA send error · Sertifi transactions for 4 cases (DM) Invalid hotel email · data returned for 3 of 4	Client Ops	done
GTM / Product1
CCA signals PRD / SF contract Existing contracts cover it ([SLDG-2373] Contract DTD: CCA delivery lifecycle events)	GTM / Product	done
Product1
Deposit-rates spike re-scope (DM) PRD changed after [SLDG-3104] Deposit-rates spike; requirements moved to SPI	Product	handed off
OBT1
Do phone cancellations reach SIG? (DM) · Contract DTD ownership Only if the supplier updates · tech leads own DTDs	OBT	done
SLDG Booking1
Hotelbeds booking stuck in NEEDS_REVIEW Needs reconciliation · [SLDG-3120] unassigned	SLDG Booking	open
Payments / GTM1
Canary wants the CVC sent with the VCC Open · no ticket	Payments / GTM	open
Data Platform1
[DPLAT-2557] Land CCA delivery events in Snowflake Needs a SAPI backfill owner	Data Platform	open
Others1
[CON-1035] Microservices codeowner review · public vs Omni monitors · Sertifi count · incident.io handle mapping Merged · kept separate · explained · mapping fix requested	Others	done

Follow-ups · 11 open of 19


CCA reseed dead letters	SLDG	SLDG-3130
Book-time exchange rates on events	SLDG	SLDG-3136
Sertifi auto-enable flag	SLDG	SLDG-3147
AVH child ages on Hotel_Sell merged Oct 6, after the handoff	SLDG	he-relay#2296SLDG-2689
Container restarts monitor standard merged Oct 6, after the handoff	SLDG	engine-search#1459SLDG-3152
Navan Sabre code-12 audit: 102 of 854 bookings, $10,016	SPI · Client Ops	SPI-483
Navan resort-fee comms	Client Ops	link
Monitor routing fixes (booking and search)	SLDG	engine-booking-api#3946
Hotelbeds reconciliation after a Nexus cancel unassigned	SLDG	SLDG-3120
Omni FA dependency monitor on an error rate PR closed unmerged; threshold only raised in Datadog, Terraform still says 100	SLDG	SLDG-3108
Sabre 10% commission default approved, not merged	SPI	he-relay#2293SPI-485
Nuitee: backfill missing resort fees from the tax-schema endpoint promised for Oct 5, no PR yet	SPI	SPI-474
CCA delivery events in Snowflake (backfill) accepted by Data Oct 1	Data Platform · SLDG	DPLAT-2557
Re-shop at confirm (Hotelbeds/WebBeds) stalled: WebBeds attempt rolled back Oct 2, Hotelbeds attempt closed	SLDG · SPI	link
AVH tax fallback at confirm waiting on Pricing's estimate	Pricing	link
FX backfill for existing Omni bookings in Salesforce asked twice by Finance; SLDG-3154 only covers future changes	SLDG	link
Keep dev smoke tests out of the no-offer P1 monitor fired 9× this week	SLDG	link
Reply to the RNPL refundability question unanswered since Oct 1; reply, no ticket needed	SLDG	link
Canary: send the CVC with the VCC on CCA creation	SLDG Booking	link

Done this week · 8 follow-ups closed, 25 tickets / PRs


CCA reseed dead letters	SLDG	SLDG-3130
Book-time exchange rates on events	SLDG	SLDG-3136
Sertifi auto-enable flag	SLDG	SLDG-3147
AVH child ages on Hotel_Sell merged Oct 6, after the handoff	SLDG	he-relay#2296SLDG-2689
Container restarts monitor standard merged Oct 6, after the handoff	SLDG	engine-search#1459SLDG-3152
Navan Sabre code-12 audit: 102 of 854 bookings, $10,016	SPI · Client Ops	SPI-483
Navan resort-fee comms	Client Ops	link
Monitor routing fixes (booking and search)	SLDG	engine-booking-api#3946


Navan stability	Session at ConfirmRate ([SPI-420] Open Amadeus sessions at ConfirmRate, [SPI-421] Open AVH sessions at ConfirmRate) · Sabre postpaid fees ([SPI-480] Sabre resort fees (code 12) postpaid) · RECHECK filter ([SLDG-3134] Search flag to drop Hotelbeds RECHECK rates, [SPI-469] Gateway drops Hotelbeds RECHECK rates) · WebBeds block ([SPI-409] Block WebBeds with the offer-token allocation) · price-shopping pass-through ([SLDG-3052] Per-supplier pass-through for partner searches) · void payment after Book cancel ([SLDG-3097] Void payment after Book cancel or timeout) · cancel idempotency ([SLDG-2859] submitCancellation idempotency)
Data & events	Refund objects ([SLDG-3116] Consistent customer and supplier refunds on events) · FX rates on events ([SLDG-3135] USD exchange rates on booking events–[SLDG-3140] Exchange rates on cancellations) · Sertifi auto-sync flag ([SLDG-3147] Sertifi auto-enable flag) · CCA dead letters ([SLDG-3130] CCA reseed dead letters) · deposit-rates spike ([SLDG-3104] Deposit-rates spike)
Alerting	[SLDG-3105] Unify lodging alerts for the Navan ramp · [SLDG-3109] incident.io-only paging · [SLDG-3106] Remove booking INTERNAL count alerts · [SLDG-3054] Drop old lodging metric names · [SLDG-3149] Integrations monitors to SPI · [SLDG-3131] Drop Wyndham from dev smoke tests · [SLDG-3133] Keep pods up during canary smoke · [SLDG-3053] Stg sandbox partner search rollout

Query shortcuts


All production confirmRate errors	Logs	`env:prd-use1-default service:engine-integrations-lodging-gateway @io-htleng-operation:confirmRate status:error`
Amadeus confirmRate and book errors	Logs	`… @io-htleng-supplier-name:amadeus @io-htleng-operation:(confirmRate OR book) status:error`
Omni confirm error rate by supplier	Metrics	`confirmRate_error ÷ confirmRate_response by {supplier}`
Every SLDG alert transition	Events	`source:alert status:(error OR warn) team:sapi-lodging -@monitor_id:(282839073 OR 305966076) -"[SPI"`
Production Search errors	Logs	`env:prd-use1-default service:(supply-api-search-lodging OR supply-api-search-partner-lodging) status:error`
Production Booking errors	Logs	`env:prd-use1-default service:(engine-booking-lodging-service OR engine-booking-lodging-jobs) status:error`
Dead-lettered tasks by env	Metrics	`max:engine_booking_tasks_dead_letter_count{…} by {env}` · dev 42 · stg 6 · prd sandbox 4 · prd 0
No-offer rate by phase (smoke-test check)	Metrics	`no_offers_sent ÷ request_count by {phase}`
Production pings	#sapi-lodging-alerts	Messages starting with `[SLDG-prd]`
Dev / staging pings	#sapi-lodging-alerts-secondary	`[SLDG-dev]`, `[SLDG-stg]`
Smoke-test traffic in production	Logs	`@io-htleng-smoke-test:true env:prd-use1-default` · add `status:error` to confirm 0 errors
All SLDG monitors	Monitors	`[SLDG-` in the name