Close 장애 기록

1. 2026년 9월 29일 오후 5:48 UTCinvestigating

   우리는 현재 가까운 애플리케이션의 부품과 관련된 문제를 해결하고 있습니다.

2. 2026년 9월 29일 오후 6:11 UTCmonitoring

   문제는 식별 및 복제되었습니다. 우리는 모든 것을 정상에 돌려보내는 신청을 계속합니다.

3. 2026년 9월 29일 오후 6:15 UTCresolved

   이 사건은 해결되었습니다.

4. 2026년 10월 5일 오후 5:25 UTCpostmortem

   우리의 서비스의 중단을 위해 근실하게 apologizes. 우리는 우리의 플랫폼의 안정성을 아주 심각하게 가지고 갑니다. 다음은 무슨 일이 있었는지 설명하고 우리가 발생하는 다른 중단을 방지하는 방법. ## 충격 닫기 시스템은 9 월 29, 17 : 40 UTC에서 9 월 29, 18 : 10 UTC로 30 분 동안 사용할 수 없습니다. ## 뿌리 원인과 해결 새로운 응용 프로그램 버전의 배포 중, 우리는 사용자 역할에 권한을 추가하고 기존의 사용자 정의 역할에 할당 된 다단계 마이그레이션을 ran. 단계 마이그레이션의 타이밍은 완전히 생각되지 않았습니다. 많은 응용 서버는 여전히 데이터베이스의 새로운 권한을 인식하지 않은 이전 버전을 실행했습니다. 결과적으로 오류율이 증가했습니다. 우리는 기존 역할에서 허가를 제거하여 서비스를 복원했습니다. 비슷한 사건을 방지하기 위해, 우리는 단계적 롤아웃을 요구하는 마이그레이션을 식별하기 위해 테스트 및 배포 파이프라인을 업데이트하고 있습니다. ## 타임라인 * 17:40 UTC - 마이그레이션이 실행되고 PR이 합병되었습니다. * 17:41 UTC - API 실패 및 응용 프로그램 응답의 엔지니어링 팀 경고 * 17:43 UTC - 엔지니어링 팀은 사건 응답, 상태 페이지 업데이트 시작 * 17:44 UTC - 잠재적 인 루트 원인 확인 \ (backfill 마이그레이션은 코드가 배포되기 전에 실행되었습니다). * 17:44 UTC - 우리는 우리가 앞으로 떨어질 수 있다고 생각하지만, 더 오래 걸릴 것을 깨달았다 그래서 우리는 그 다음 마이그레이션을 반전하기로 결정했다. 이 정전 시간에 추가. * 18:10 UTC - 엔지니어링 팀은 기존의 사용자 정의 역할에서 새로운 권한을 삭제하는 스크립트를 실행 * 18:10 UTC - 응용 프로그램 로딩 다시, 오류율 떨어졌다. 고객 충격 끝.

공식 인시던트 업데이트를 자동 번역했습니다.

1. 2026년 8월 11일 오후 1:12 UTCinvestigating

   We are again investigating an elevated number of database failures.

2. 2026년 8월 11일 오후 2:15 UTCinvestigating

   We are continuing to investigate the issue.

3. 2026년 8월 11일 오후 2:53 UTCmonitoring

   We have identified the problem, adjusted our scaling strategy, and are seeing healthy recovery of all systems.

4. 2026년 8월 11일 오후 3:15 UTCresolved

   This incident has been resolved.

5. 2026년 8월 13일 오후 10:56 UTCpostmortem

   ## Summary & Impact Close sincerely apologizes for the interruption of our service. We take the stability of our platform very seriously, and this incident is not acceptable to us. Below is an explanation of what happened and how we will prevent another such interruption from occurring. On August 11th, 2026, Close was intermittently unavailable for a total of about 2.5 hours between 08:00 UTC and 17:03 UTC. During those periods, the Close web app, mobile apps, and API returned errors or failed to load, affecting customers in all regions. Because the disruption was intermittent rather than continuous, many customers likely experienced it as a longer period of unreliability across the working day rather than as a handful of discrete outages. There was also added confusion because we marked our status page "resolved" twice during the day, at 09:26 UTC and again at 15:15 UTC, before the underlying cause had been fixed. We are improving our process so that we do not mark an incident as resolved until we have a good understanding of the situation and applied appropriate mitigations, not just because the symptoms have subsided. ## What went wrong? A routine code change to our email and calendar syncing service altered the way it looks up account information in our database. The new version looked equivalent to the code it replaced, but it caused our database to perform dramatically more internal work for each lookup, enough to exhaust the memory on our primary database server and repeatedly crash it. Because each crash was followed by an automatic recovery, the pattern looked like general instability rather than the symptom of a specific change, which is why it took us longer than we would have wished to trace it back and revert it. ## How we’ll fix this We have already reverted the problematic change and made configuration adjustments to our database hosts to reduce the severity of this class of failure. Close is currently stable. Going forward, we will improve the design of our Email & Calendar Syncing service and similar services which touch Organizations, Users, and Memberships. We will also improve our query performance visibility and alerting, so that slow or inefficient queries and new, unexpected query patterns are identified sooner than they were here. ## Technical Details For those interested in the specifics, here is the full sequence of events and the underlying database behavior. ### Timeline \(all times UTC\) * **August 10, 20:34** — The change finishes deploying to production. Our primary database shard briefly becomes unstable and fails over to a redundant node, producing a few minutes of errors. Because the change deployed after peak usage and deployment hours, traffic subsided shortly afterward and the database remained stable overnight. * **August 11, 08:00** — As application usage and engineering deployments begin increasing, error rates on our primary database shard start to climb. * **08:17** — Our monitoring alerts fire. * **08:26** — Close Engineering is investigating and has posted a status page incident. * **08:46** — We restart the affected database processes, allowing the system to recover. * **09:26** — All Close systems have returned to what looked like normal operation, and we mark the status page resolved. * **12:11 – 17:03** — The same loop of elevated errors, instability in the primary shard, failover to a redundant node, and temporary recovery repeats itself several times. * **14:45** — We find that the database is exhausting its allowed number of open files and raise that limit. This improves stability but does not address the underlying cause. * **15:15** — Systems appear to have recovered again, and we mark the incident resolved a second time. * **15:50** — We identify the previous day's deployment as the likely culprit. * **16:51 – 17:18** — We begin reverting the change when the database becomes unstable again. Customer-facing errors subside at 17:03, and the rollback completes at 17:18. ### The underlying database behavior Our Email & Calendar syncing service regularly queries a database for an up-to-date state of Organizations, Users, and their active Memberships. We store this information in our MongoDB cluster, on its primary shard. We were refactoring some of this code as part of normal tech debt work to bring older code in line with our current patterns. Previously, this query used a `$in: […long list of active organizations' IDs…]` clause as the filter for this query. This wasn’t necessarily very efficient, but it was acceptable because the query ran at a relatively low rate \(only happened every 50 seconds for one syncing process, of which we have 120-180 running depending on the demand\) and runs asynchronously. The refactor changed that query to find applicable Memberships via a `$or: [{organization: someOrgId, user: someUserId}, …long list of such conditions for all users with email/calendar accounts…]` clause. This at first looked like it should have performance roughly equal to the previous query. However, we have since discovered that, while our MongoDB cluster handled large `$in` queries decently well, a large set of compound queries all put into a root-level `$or` clause had disastrous performance characteristics. Even ~3-4 such queries per second, all running roughly around the same time alongside other normal requests, could overwhelm even our heavily scaled database node. Our current understanding, based on our review of the MongoDB source code and ad hoc testing, is that the issue relates to how MongoDB plans large root-level `$or` queries. MongoDB may evaluate the clauses of this query shape as separate sub-plans rather than as a single index scan. In our case, a query could contain approximately 50,000 clauses. At three such queries per second, the database was being presented with roughly 150,000 `$or` clauses per second. We believe this generated tens of thousands of index scans and WiredTiger cursors, churned the query-plan cache, and placed extreme pressure on memory and other database resources. This behavior is consistent with the out-of-memory terminations, file-handle exhaustion, and other low-level failures we observed. Reverting the new query shape eliminated the resource spikes and restored database stability. Pinpointing the query responsible for the database load took longer than anticipated because the seemingly small query rate, a flood of other slow queries, and an immediate database recovery obscured the underlying cause.

1. 2026년 8월 11일 오전 8:26 UTCinvestigating

   We are currently investigating an elevated number of database failures.

2. 2026년 8월 11일 오전 8:54 UTCidentified

   We have identified an issue impacting our production database. Close Engineering is taking action to restore functionality. Users should notice improved performance over the next several minutes.

3. 2026년 8월 11일 오전 9:04 UTCmonitoring

   We have remediated the issue with our production database. Users should be able to use the Close app normally. Close Engineering is monitoring system performance to confirm resolution.

4. 2026년 8월 11일 오전 9:26 UTCresolved

   This incident has been resolved.

1. 2026년 6월 29일 오후 5:46 UTCinvestigating

   We have become aware of degraded processing of background tasks in the Close system. We are currently investigating the issue.

2. 2026년 6월 29일 오후 7:25 UTCmonitoring

   We have identified the root cause and deployed a remediation. Background task processing performance is improving. Some customers may still see delayed processing of background tasks. Close Engineering is monitoring for improved performance.

3. 2026년 6월 29일 오후 7:39 UTCresolved

   This incident has been resolved. Background tasks are processing normally.

1. 2026년 6월 3일 오후 2:50 UTCinvestigating

   Changes done to CRM data are reflected in search with a delay.

2. 2026년 6월 3일 오후 3:21 UTCinvestigating

   We are continuing to investigate the issue.

3. 2026년 6월 3일 오후 4:41 UTCinvestigating

   We are continuing to investigate the issue. Updates to Close made via the web app and API may take longer than normal to appear in search results.

4. 2026년 6월 3일 오후 5:45 UTCidentified

   We have implemented a remediation for the delayed search indexing and are monitoring for improvement. Updates to Close made via the web app and API may take longer than normal to appear in search results.

5. 2026년 6월 3일 오후 6:17 UTCmonitoring

   A fix has been implemented and we are monitoring the results. Updates to Close made via the web app and API may take longer than normal to appear in search results.

6. 2026년 6월 3일 오후 6:53 UTCresolved

   This incident has been resolved. The Close app and API are now functioning normally.

1. 2026년 5월 12일 오전 9:34 UTCinvestigating

   Some users are experiencing connection issues and application errors. Our engineers are currently investigating issues loading the Close application, making & receiving calls, and responding to Close API requests.

2. 2026년 5월 12일 오전 11:07 UTCmonitoring

   A fix has been implemented and we are monitoring the results.

3. 2026년 5월 12일 오전 11:17 UTCresolved

   This incident has been resolved.

1. 2026년 4월 23일 오후 4:56 UTCinvestigating

   We are currently investigating this issue.

2. 2026년 4월 23일 오후 5:14 UTCidentified

   The issue has been identified and the fix is being implemented.

3. 2026년 4월 23일 오후 5:56 UTCmonitoring

   A fix has been implemented and we are monitoring the results.

4. 2026년 4월 23일 오후 6:50 UTCmonitoring

   We've deployed a fix, we're continuing to see improvements in overall system performance, we expect recovery slowly over the next several minutes.

5. 2026년 4월 23일 오후 7:44 UTCresolved

   This incident has been resolved.

1. 2026년 4월 23일 오후 4:16 UTCinvestigating

   We are currently investigating this issue.

2. 2026년 4월 23일 오후 4:56 UTCinvestigating

   We are experiencing an issue with a 3rd party provider.

3. 2026년 4월 23일 오후 5:32 UTCmonitoring

   We are experiencing an issue with a 3rd party provider.

4. 2026년 4월 23일 오후 7:45 UTCmonitoring

   We're continuing to monitor the situation and expect gradual recovery over the next hour.

5. 2026년 4월 23일 오후 9:18 UTCresolved

   This incident has been resolved.

1. 2026년 3월 16일 오후 4:08 UTCinvestigating

   We are currently investigating this issue.

2. 2026년 3월 16일 오후 4:16 UTCinvestigating

   We are investigating the issue and seeing some recovery.

3. 2026년 3월 16일 오후 4:27 UTCmonitoring

   Database infrastructure is operational.

4. 2026년 3월 16일 오후 4:45 UTCmonitoring

   Close services continue to recover. Most customers should see normal performance. Close Engineering continues to monitor the recovery.

5. 2026년 3월 16일 오후 4:55 UTCresolved

   This incident is resolved. Close has returned to normal performance.

6. 2026년 3월 17일 오후 3:13 UTCpostmortem

   # Summary Close sincerely apologizes for the interruption of our service. We take the stability of our platform very seriously. Below is an explanation of what happened and how we will prevent another such interruption from occurring. The Close App and API were severely disrupted for all users for 18 minutes between 15:57 and 16:15 UTC on Monday March 16, 2026. Automated systems restored functionality by 16:15 UTC. # Incident Timeline | Time \(UTC\) | Event | | --- | --- | | 2026-03-16 | | | 15:57:00 | **Automated Database Maintenance begins** | | 15:58:42 | **Incident detected by monitoring** | | 15:58:49 | **Escalation acknowledged by Close Engineering On-Call** ​ | | 16:02:41 | **Incident accepted by Close Engineering** ​ | | 16:15:00 | **Affected database restored to normal operation** | | 17:09:22 | **Incident resolved** | # Root Cause and Resolution At ~15:57 UTC an automated database maintenance process began that starved our production database cluster of resources. This caused our primary instance and both replica instances of the production cluster to become unavailable for ~18 minutes. The system recovered and began performing normally at 16:15 UTC. To prevent this from happening in the future Close Engineering has provisioned additional resources to the affected database cluster to make it more resilient to sudden demand spikes. Close Engineering will also perform an audit of the maintenance process' configuration to identify opportunities to reduce its impact on system performance.

1. 2025년 11월 5일 오후 6:42 UTCinvestigating

   We are receiving an increased number of reports that users are unable to use calling when using the Desktop Application. Calling within the browser works fine. We're investigating.

2. 2025년 11월 5일 오후 7:21 UTCidentified

   We have identified the issue and are actively working on a fix.

3. 2025년 11월 5일 오후 7:56 UTCresolved

   The issue has been resolved. Please reach out to [email protected] if you continue to have calling issues after refreshing.

1. 2025년 11월 1일 오전 8:55 UTCinvestigating

   We're currently investigating a slowdown in application performance. You may experience delays in loading pages. The engineering team is currently looking into resolving this as quickly as possible.

2. 2025년 11월 1일 오전 10:32 UTCidentified

   The issue has been identified and a fix is being implemented.

3. 2025년 11월 1일 오전 10:53 UTCresolved

   This incident has been resolved.

1. 2025년 10월 23일 오전 11:40 UTCmonitoring

   We identified an issue causing Lead Pages to load slowly during Dialer sessions. A fix has been implemented and will be deployed shortly. After deployment, we’ll monitor the system to ensure the issue is fully resolved. New updates to follow.

2. 2025년 10월 23일 오후 3:52 UTCresolved

   This incident has been resolved.

3. 2025년 10월 27일 오후 4:51 UTCpostmortem

   Close sincerely apologizes for the interruption of our service. We take the stability of our platform very seriously. Below is an explanation of what happened and how we will prevent another such interruption from occurring. ### Impact Between Tuesday 10/21 14:46–17:24 UTC, Wednesday 10/22 7:50–17:00 UTC, and Thursday 10/23 8:00–15:00 UTC, customers relying on real-time updates in Close \(which is critical for proper & timely functioning of our calling features and the Dialer\) experienced delays, outdated data being displayed on their screens, Dialer disconnections, and other unexpected behavior. ### Root Cause and Resolution On Monday 10/20 at 20:09 UTC, a subtle bug was introduced, which resulted in UI clients slowly opening more and more WebSocket subscriptions for our Data Monitoring Service – a component which powers real-time updates for our Activity Reporting pages. Over the next few days, the number of subscriptions kept rising. The rise was slow enough where it didn’t trigger our alerting, but fast enough to eventually start putting a real strain on our infrastructure and saturating our processing capacity. Our first course of action was to scale the supply to meet the demand, but the demand continued to rise more quickly. Because of unrelated rate-limiting changes that looked more directly responsible for the limitation of the supply, our engineering efforts went in a few wrong directions until finally the root cause – the UI bug – was identified and fixed on Thu 10/23 at 5:19 UTC. However, because the fix required clients to reload their UI, intermittent problems continued until enough clients have reloaded, finally resolving the problem on Thu 10/23 at 15:00 UTC. To prevent this problem from ever happening again, we are improving our monitoring, alerting, testing, and in the longer-term the architecture of our WebSocket subscription service. ### Timeline * Mon 10/20 20:09 UTC: A subtle bug is introduced and over the next few days keeps slowly increasing the number of WebSocket-based Data Monitoring Service subscriptions. * Tue 10/21 14:46 UTC: Close receives first signals of Dialer problems where Users hear the callee many seconds before the corresponding Lead Page loads and immediately start looking into the problem. * Tue 10/21 5:24 UTC: First fixes aimed at increasing the allowed rate of processing of the system land in production. The situation appears resolved. * Wed 10/22 7:50 UTC: The problem comes back despite the fixes. Engineers increase the scale of the system to meet the increasing demand, but the demand keeps saturating the scale. Status page is opened at 10:35 UTC \([https://status.close.com/incidents/ydzwc4l6xmhc](https://status.close.com/incidents/ydzwc4l6xmhc)\). Problems continue on and off until they fully subside around 17:00 UTC * Thu 10/23 5:19 UTC: Engineers identify the root cause and fix it. * Thu 10/23 8:00 UTC: Concerning metrics continue to emerge despite the fix. Customer reports follow and Status page is opened at 11:40 UTC \([https://status.close.com/incidents/j49qgd3m049n](https://status.close.com/incidents/j49qgd3m049n)\). Engineers continue working on the impact mitigation and root cause diagnosis until it is found and fixed at 11:47 UTC \(excessive WebSocket Subscriptions for our real-time Data Monitoring Service, coming from UI clients who visited the Activity Reporting page\). However, the fix being picked up requires some idle UI clients to reload, hence the saturated demand continues until 15 UTC despite scaling attempts. * Thu 10/23 15:00 UTC: The problem is officially resolved.

1. 2025년 10월 22일 오전 10:35 UTCinvestigating

   We are currently investigating problems with Lead Pages being loaded too slowly while on a Dialer session.

2. 2025년 10월 22일 오전 11:10 UTCmonitoring

   A fix has been implemented, and we are currently monitoring this. New updates to follow.

3. 2025년 10월 22일 오전 11:50 UTCresolved

   This incident has been resolved.

1. 2025년 10월 20일 오후 2:20 UTCinvestigating

   There’s currently an outage on AWS’s end in a certain region. While our hosting is mainly in another region, many other services, including calling, are affected.

2. 2025년 10월 20일 오후 8:27 UTCresolved

   This incident has been resolved.

1. 2025년 10월 20일 오전 10:38 UTCmonitoring

   There’s currently an outage on AWS’s end in the us-east-1 region. While our hosting is mainly in another region, many other services, including calling, are affected. We are actively monitoring this, and we will keep you updated

2. 2025년 10월 20일 오전 11:16 UTCmonitoring

   AWS has reported partial recovery on this incident (https://health.aws.amazon.com/health/status), but services have yet to recover fully. Our service provider, Twilio, is also actively monitoring this (https://status.twilio.com/) We will continue to keep you posted.

3. 2025년 10월 20일 오후 1:12 UTCresolved

   This incident has been resolved.

1. 2025년 10월 14일 오후 3:26 UTCinvestigating

   Some users are experience downtime. Our engineers are currently investigating issues loading the Close application and pulling data from the Close API.

2. 2025년 10월 14일 오후 3:48 UTCinvestigating

   We are currently applying fixes to restore operations and monitoring their impact.

3. 2025년 10월 14일 오후 3:53 UTCmonitoring

   A fix has been implemented and we are monitoring the results.

4. 2025년 10월 14일 오후 4:00 UTCresolved

   This incident has been resolved.

5. 2025년 10월 16일 오후 6:41 UTCpostmortem

   Close sincerely apologizes for the interruption of our service. We take the stability of our platform very seriously. Below is an explanation of what happened and how we will prevent another such interruption from occurring. ## Impact A small number of Close API requests began failing on October 14, 2025 for 67 minutes from 13:20Z to 14:27Z. A larger number of failures occurred for 88 minutes from 14:27Z to 15:55Z. The failures impacted our API as well as various components of the Close App. ## Root Cause and Resolution Unfortunately, a cascading set of events contributed to this failure. Three Pull Requests \(PR\) were in flight during this incident. The first and third had defects that impacted our API. Additionally, although it had no defects, the second PR added to the complexity of troubleshooting the core issues caused by the other two. The first PR made changes to our billing logic that inadvertently started making more requests than expected to an external provider. This extra load on the external provider caused a small number of our API requests to timeout. The larger impact was caused by the third PR which made changes to how some parts of our web requests are handled. We have recently updated our dependencies and are in the process of evolving various parts of our codebase to be better aligned with modern async processing techniques. This update appeared fine during normal testing but failed under our production load. The changes caused a very high utilization of our web workers’ thread pools. This in turn caused some of them to become saturated and left in a state where they were unable to process any new requests. This caused a significant number of API requests to time out. It took some time to identify which update was causing the API timeouts and revert its changes. A flaky test further delayed our deployment process. The recovery was prolonged because the failure affected our infrastructure's health check system. The same endpoints that were timing out were needed to validate and control our deployment changes. We are still reviewing all details related to this incident and are taking several actions to improve our response to complex incidents such as these. We are working to ensure the right resources are available to quickly diagnose issues, strengthening our CI/CD processes, and enhancing our internal QA procedures to better catch these complex defects during testing. ## Timeline * 13:20 UTC - Billing PR merged earlier in the morning starts causing a small number of API failures * 13:25 UTC - Engineering team alerted of API failures, response starts * 13:37 UTC - Async PR is merged, fails to deploy due to flaky e2e test * 13:45 UTC - Engineering team alerted of failing CICD deployments * 13:51 UTC - Revert billing PR, deploy fails due to flaky e2e test * 14:04 UTC - Unrelated PR merged which includes the billing change revert * 14:20 UTC - Both PR’s changes deployed * 14:27 UTC - Large spike of API failures followed up with pods failing to start due to status endpoint failing * 14:33 UTC - Start of main customer impact of the Async PR * 15:18 UTC - Revert Async PR * 15:55 UTC - Customer impact ends

1. 2025년 9월 24일 오후 12:01 UTCinvestigating

   Some customers in the EMEA region are experiencing connectivity issues to the Close app. We believe this is a widespread issue and are continuing to investigate.

2. 2025년 9월 24일 오후 12:22 UTCmonitoring

   The issue appears to have resolved itself. We are still monitoring and waiting for our Cloud provider to confirm resolution.

3. 2025년 9월 24일 오후 1:23 UTCmonitoring

   AWS confirmed they were investigating an issue but haven't yet notified us that it has been resolved. We are still monitoring for new issues.

4. 2025년 9월 24일 오후 1:31 UTCresolved

   Connectivity issues appear to have been resolved. Still waiting for AWS to confirm what actions were taken to resolve the issue.

5. 2025년 9월 25일 오전 10:19 UTCpostmortem

   Our cloud provider was able to confirm they were having issues with an external network provider during this time period that caused connectivity issues between Europe and the us-west-2 region. This aligns with the issues that were reported to our support team. We will review our monitoring and alerting processes to determine if we can be alerted sooner to these types of regional cloud connectivity issues so that we can more quickly notify our customers.

1. 2025년 9월 5일 오전 10:45 UTCinvestigating

   We are currently investigating a delay in processing triggers that kick off Workflow actions.

2. 2025년 9월 5일 오전 11:01 UTCidentified

   We have identified the problem as coming from our compute provider, AWS. An unplanned patch to our Event Stream is currently underway.

3. 2025년 9월 5일 오전 11:13 UTCresolved

   Unplanned patches have been applies and our system has processed through the delayed events.

1. 2025년 8월 6일 오전 8:16 UTCidentified

   We've identified a problem with our telephony provider, Twilio, which incorrectly flags some (not all) accounts as compromised and thus disables their ability to make calls. We have escalated the problem and are working urgently with Twilio to resolve the issue.

2. 2025년 8월 6일 오전 10:55 UTCmonitoring

   Our telephony provider, Twilio, has fixed the accounts affected so far and is actively working on preventing the problem from reoccurring.

3. 2025년 8월 6일 오후 1:08 UTCresolved

   This incident has been resolved.

1. 2025년 7월 22일 오후 7:34 UTCmonitoring

   Some customers may experience delayed updates to derived fields such as "Last Communication Date". The delay should be resolved soon.

2. 2025년 7월 22일 오후 8:18 UTCresolved

   This incident has been resolved.