Cases12 cases

2 to 6 of 12

1. IL-0572 About 10 million Pub/Sub messages backed up and Merpay accounting data was not written to Cloud Spanner for over a day

   Mercari2026

   Symptoms

   From about 6:00 on the last business day of a month, Cloud Pub/Sub messages of the Merpay accounting event pipeline began to back up, new accounting data stopped being registered in Cloud Spanner, and the accounting system intermittently could not accept new data. The backlog grew to about 10 million messages.

   Throughput degradationDatabase overloadAutoscaler misbehaviour

   Trigger

   More than a month earlier the Cloud Functions of the pipeline had been moved from 1st gen to 2nd gen (based on Cloud Run) without setting --max-instances in the deploy command, which lowered the scale limit from effectively unlimited, the 1st gen default, to 100, the 2nd gen default. The month-end peak, which on that day continued for about 40 minutes, reached that limit.

   Root cause

   Breaking change in an upgrade

   Component

   Serverless function platformMessage brokerDatabase

   Contributing factor

   Unreviewed defaultsUnknown limitChange with a delayed effectMonitoring blind spot

   Remediation

   • Re-ingested the messages saved to GCS by the Dataflow subscription, avoiding data loss

   Action items

   • Set max-instances and the other scaling parameters of Cloud Run explicitly in the deploy definition, raising max-instances from 100 to 1000 in steps
   • Set min-instances to 1 to avoid cold starts, raise concurrency from 1 to 10, and set CPU to 1 and memory from 512MB to 1Gi
   • Use the dual CPU scaling mode of spanner-autoscaler v0.8.0 with a new Total CPU threshold of 70%
   • Lower the High priority CPU threshold from 60% to 55% so scale-up starts earlier
   • Forbid Spanner scale-down between 5:00 and 7:00
   • Reserve Spanner processing units on a schedule for periods with predictable load
   • Alert in Datadog when the Cloud Run instance count reaches 40% and 60% of its limit
   • Alert in Datadog when undelivered Pub/Sub messages exceed a threshold, lowering it step by step

   Lessons

   When moving to a new generation of a managed service, defaults that the team never changed can surface as problems under load, so parameters that only matter at peak, such as scale limits, concurrency and timeouts, should be set explicitly at every migration. An outage rarely has one cause; the fact that the previous month passed under the same conditions pointed to traffic continuity, and prevention needs to look at the trigger, the amplification and what turned it into damage. Capacity settings and the monitoring that shows when they are approached have to be designed together, and teams that run systems they inherited should review these settings and assumptions regularly.

   View case
2. IL-0288 ZooKeeperWait rose to thousands of seconds on three redundant ClickHouse instances, degrading a CDN77 data pipeline

   CDN772026

   Symptoms

   At 13:30 UTC on 9 March 2026, CDN77's data team was alerted to a sudden rise in the ZooKeeperWait metric on the ClickHouse cluster behind one of its large data pipelines. Alerting for this cluster size is set at 10 sec for the low watermark and 20 sec for the high one, and the metric reached thousands of seconds. The same rise appeared on three symmetrically redundant ClickHouse instances in different datacenters, each with its own ZooKeeper instance, and the pipeline stayed degraded for over an hour, with no observable impact on internal users, customers or partners.

   Replication lagThroughput degradation

   Trigger

   From 13:15 UTC two client machines that had been cut off from the network for several months came back and their log shippers sent months of queued, very small batches of health check data with old timestamps, spanning a very large time range.

   Root cause

   Capacity exhaustion

   Component

   DatabaseCoordination and key-value storeDatabase replicationTelemetry agent

   Contributing factor

   Monitoring blind spotUnreviewed defaultsIgnored early signal

   Detection

   An alert on the ZooKeeperWait metric fired at 13:30 UTC, fifteen minutes after the problematic inserts began at 13:15. The cluster's alerting uses a low watermark of 10 sec and a high watermark of 20 sec.

   Remediation

   • Dropped and recreated the one table partitioned by 5 minutes that had not cleared its backlog of parts
   • Broadened that table's partitioning from minutes to hours

   Action items

   • Reject out-of-range timestamps in the root MaterializedView tables
   • Record statistics on rejected rows in a second materialized view
   • Refine the deliberately wide rejection boundaries as data on rejections comes in
   • Keep a minimal test scenario with an insert query whose timestamps span days or months
   • Make a range of observability improvements

   Lessons

   Data structure matters even in a pipeline built around volume: incoherent timestamps can destabilise it as effectively as changes in cardinality. A metric that looks odd and cannot be explained at once should not be set aside under incident pressure; the falling mean insert size was what led to the cause. Means and medians each have their place, and here the mean, not the median, exposed the problem. Partitioning has to be chosen carefully: too few partitions give oversized parts, slow mutations and costly or ineffective TTL, while too many slow column scans, add final merging to aggregating queries and overload the Keeper.

   View case
3. IL-0390 PostHog US Cloud event ingestion slowed to delays of up to 2 days as writes to the Persons Postgres table stalled

   PostHog2025

   Symptoms

   From 11 November 2025 at 16:02 UTC, PostHog's event ingestion pipeline for US Cloud became severely degraded, and events for all customers in the US region were processed with delays from 10 minutes up to 2 days. Writes to the Postgres database that stores Person information could take multiple seconds per update, and the pipeline slowed until no standard scaling option helped.

   Throughput degradationHigh latency

   Trigger

   The 32-bit OID space used to track TOASTed values of the Persons table's properties column, a JSONb field that often exceeds the 2KB threshold for out-of-line storage, was nearly exhausted.

   Root cause

   Capacity exhaustion

   Component

   DatabaseMessage broker

   Contributing factor

   Unknown limitMonitoring blind spotNo runbook

   Remediation

   • Created a new partitioned Persons table with a fresh TOAST table and OID space
   • Added database triggers to keep the old and new Persons tables in sync for live writes
   • Copied the existing Person data from the old table to the new one with a backfill script
   • Changed ingestion and application code to write to the new table and read from it first, with fallback to the old table
   • Deployed the web app changes to production under manual control rather than a fully automated rollout
   • Scaled ingestion workers to process the accumulated backlog
   • Paused ingestion when the MSK local disk reached about 85%
   • Reduced the local retention of the affected Kafka topic
   • Resumed ingestion after confirming there was enough disk headroom
   • Moved the backfill into Dagster to finish it over the weekend

   Action items

   • Add metrics and alerts for heavy disk read amplification per write
   • Add metrics and alerts for TOAST table statistics
   • Add metrics and alerts for other Postgres engine limits that matter at large scale
   • Write runbooks for engine-level Postgres limits, with decision trees for migrating versus repairing in place
   • Create customer communication runbooks with a clear escalation path and redundancies
   • Evaluate other data stores for the persons database
   • Complete the reconsolidation on the new Persons tables
   • Improve the backfill tooling and the safeguards for Kafka tiered storage
   • Design and operate Person-like tables so they avoid similar limits

   Lessons

   TOAST OID exhaustion degrades writes long before the hard limit of about 4 billion OIDs, it is rare even at large scale, and standard dashboards for CPU, memory, IOPS and lock contention did not immediately point at it, so diagnosis took a day and a half without monitoring of TOAST growth, OID usage or disk reads per write. Many core features depended on timely updates to the persons table, so its slowdown reached a wide part of the product. The first backfill approach lacked visibility for a long migration, the Kafka tiered storage settings and alerts were not tuned for a large backlog, and the people who normally handle customer communications were all unavailable.

   View case
4. IL-0582 Jobs piled up in Kanmu's PostgreSQL-based qg job queue on Aurora as dequeue slowed and dead tuples kept growing

   Kanmu2024

   Symptoms

   A Datadog alert fired because a large number of unprocessed jobs had piled up in Kanmu's job queue and were not being worked off. The jobs were the batch enqueued at about the same time every day, in the usual volume, but the workers took far longer than normal to dequeue and process them.

   Throughput degradationHigh latency

   Trigger

   A heavy ad hoc analysis query from the BI tool Redash ran for a long time on the writer instance, because the Redash data source pointed at the cluster endpoint.

   Root cause

   Operator error

   Component

   DatabasePostgreSQL vacuum

   Contributing factor

   Change with a delayed effectUnreviewed defaults

   Detection

   A Datadog alert on unprocessed jobs piling up in the job queue.

   Remediation

   • Changed the Redash data source to the reader endpoint
   • Ended the long-running Redash query with pg_terminate_backend
   • Ran VACUUM on que_jobs by hand again, which cleaned up the dead tuples

   Action items

   • Make Redash unable to connect to the writer instance even if the writer endpoint is set, by mapping its host name to a TEST-NET-1 address in the Redash container's /etc/hosts
   • Set statement_timeout on PostgreSQL users so that long queries stop automatically
   • Consider improving qg's dequeue algorithm or adopting a job queue such as river that uses SELECT FOR UPDATE SKIP LOCKED

   Lessons

   A long-running query on the same database instance, even one unrelated to que_jobs, can keep VACUUM from removing que_jobs dead tuples, so the table keeps growing and the job queue slows down. qg's design with advisory locks and a recursive CTE is sensitive to dead tuples.

   View case
5. IL-0447 An every RDS for MySQL instance slowed badly when its gp2 EBS BurstBalance ran out and IOPS fell to the baseline

   every2024

   Symptoms

   The BurstBalance of an RDS instance began to fall rapidly and finally ran out, and the database's performance dropped sharply. The instance ran MySQL 5.7.44 on db.t3.medium with general purpose SSD (gp2) storage of 30GiB.

   Database overloadThroughput degradation

   Trigger

   A complex SELECT query began writing large amounts of data to temporary tables, which raised WriteIOPS and used up the burst credits.

   Root cause

   Capacity exhaustion

   Component

   DatabaseCloud block storage

   Contributing factor

   Inefficient query

   Action items

   • Consider tuning the query by reviewing indexes and rewriting it
   • Consider moving the storage from gp2 to gp3 or increasing its size to raise the baseline IOPS
   • Consider adjusting tmp_table_size and max_heap_table_size
   • Consider moving to a more powerful instance type
   • Consider monitoring BurstBalance with CloudWatch alerts and adding external monitoring such as Pingdom

   Lessons

   A complex query can have a large effect on I/O, so database design and queries need continuous improvement, and query performance needs regular monitoring.

   View case

By root cause4