Live outage feed

Outages and recoveries from the services you depend on.

Earlier incidents

40 incidents

Back to latest

Oct 12, 2020

  1. RunscopeAPI Monitoring slowness

    Our engineering team has completed the fixes to resolve this incident. They are continuing to make additional optimizations but API Monitoring should be fully functional. After we finish reviewing the incident, we will post a postmortem report here. We apologize sincerely for the disruption. Please note, status.runscope.com will be migrating to status.blazemeter.com in the near future. If you have already subscribed to notifications here, we will copy them over automatically.

    Resolved

Oct 9, 2020

Oct 6, 2020

Oct 1, 2020

  1. Travis CIBuilds are not triggering for travis-ci.com

    We are extremely sorry about the outage on this incident and we would like to share some action that has been taken; * Issue reported date: The issue has been reported on 24.09.2020 09:12 UTC. * Issue: **Unstable VCS service, high memory usage** * Reason: Issue in Bitbucket implementation, infinity loop when a user has been suspended or without permissions to the repository * Solution: Check suspended flag before API calls, infinite loop retries removed. * Fixed time: 24.09.2020 10:14 UTC * Action required from users: None * Issue status: Fixed Thank you for your understanding.

    Resolved

Sep 30, 2020

Sep 29, 2020

Sep 15, 2020

Sep 1, 2020

Aug 18, 2020

  1. HoneybadgerAPI downtime

    We had an automation failure that caused a few minutes of downtime for our ingestion API. We are reworking our automation to avoid this problem in the future.

    Resolved

Aug 13, 2020

Aug 12, 2020

Jul 28, 2020

  1. StreamIncreased API latency

    AWS Networking issue is now resolved. We are now cleaning up our temporary remediations since they are not needed anymore. Traffic is back to normal for the last hour.

    Resolved

Jul 27, 2020

  1. DopplerUsers are unable to connect to Doppler servers

    Our first critical outage occurred today, and it led to almost 2 hours of downtime but thankfully no data loss. This was an unacceptable amount of downtime and could have been completely prevented. We take your trust very seriously as Doppler is a critical path in your devops and productivity workflows. We have learned greatly from this experience, from fixing the root cause and adding checks to prevent this kind of outage in the future. Here is what happened: # Timeline **March 6th, 2019 - 3:04 PM \(PST\)** [Heroku](https://heroku.com) starts an automated maintenance on our primary postgres database. This process includes creating a follower of our primary database on the newest postgres version, then hard forking the database and setting it as the new primary database. After the primary database is in use, the old database is removed. We were warned about this migration ahead of time and assumed the maintenance/migration would be quick and all of the credentials would automatically be updated across our environments. **March 6th, 2019 - 3:20 PM \(PST\)** As our servers start crashing and we receive a flood of [Bugsnag](https://bugsnag.com) error reports. Digging through the stack traces and logging, we realize a failure occurred during the migration where our primary database url was revoked but the new primary database url had not been automatically set as an environment variable. This made our servers attempt connection to an invalid database url which caused our servers to crash. This same failure affected our failover servers which relied on a follower database of our primary database. Since Doppler relies on Doppler for environment variables, it created a circular loop where we need to be up to boot up. Removing Doppler as a source for environment variables is tricky, as we only stored our environment variables in Doppler. Those environment variables are tokenized with our security vendor so it was impossible for us to copy them from our database into Heroku manually. Instead, we had to go through every service we used to grab or create new credentials through each of their individual dashboards. **March 6th, 2019 - 3:30 PM \(PST\)** We initiated our recovery plan: 1\. Report the incident on StatusPage 2\. Put a list together of all the credentials needed 3\. Find correct database url 4\. Modify our environment variables with credentials 5\. Deploy new code with our up-to-date environment variables **March 6th, 2019 - 3:50 PM \(PST\)** After finding all of our credentials / environment variables we needed, we started inputting them manually into Heroku environments. This had to be done across all of our services for Doppler to work properly. After setting most of our environment variables, we ran into a critical problem. Heroku would not let us set the new database url because it was a managed variable controlled by Heroku Postgres. In addition, our servers started scaling up rapidly as they were constantly crashing and rebooting, each time hitting our API that was already down. This endless cycle created timeouts and produced a massive number of logs, making it very hard to debug. To counteract this, we tried to disable autoscaling and scaled our servers down to 1 dyno. This process kept returning an error as Heroku would mark it as an invalid request. After 10 minutes, we were able to successfully scale our servers down to 1 dyno. Now that logging is easier to comb through, we refocused on fixing our postgres invalid credentials problem. We found a buried option in a submenu of the Managed Postgres Dashboard to force a rotation of our database credentials. Next, Heroku immediately propagated those new credentials to our environment variables as we expected. **March 6th, 2019 - 4:15 PM \(PST\)** Now that all of our correct environment variables are on Heroku, we started step 5. Though the Doppler API was down, our SDK was able to fallback to the environment variables set on Heroku. Quickly after deploying the fix and our website is back up, we realize our security vendor's API url is incorrect from the one on the dashboard, as they had created a dedicated one for us. **March 6th, 2019 - 4:30 PM \(PST\)** After digging through a hundred old slack messages and talking with support, we found the dedicated API url to use. Now our servers are fully back up and running. **March 6th, 2019 - 4:50 PM \(PST\)** After 20 minutes of additional monitoring and stress testing, the StatusPage is updated with the indication that the incident is resolved. # What We Learned ### **Circular Loop** Doppler being a critical path for Doppler is a linchpin waiting to be pulled. We needed to have a fallback option built-into our systems so that Doppler can boot up after a full outage. This will be done by ensuring Heroku environment variables are always up to date. ### **Stronger Prechecks** Previous to the outage, our precheck script strictly checked for the presence of all our required environment variables. This is not good enough. Our new precheck script also initializes all libraries and clients \(including our database client\) to ensure all the credentials are correct. During the outage, when our database url was revoked, the old prechecks still passed as a url was present. It now ensures authorization ahead of time and done successfully. ### **DevOps Expertise** A wake up call that we will not be able to run on Heroku for too long. Heroku provides a lot of value for the price tag but comes at a deep expense: lack of devops expertise. Overtime we will migrate to AWS and build out our own devops workflows that can support our extremely high SLA and fault tolerant requirements.

    Resolved

  2. DopplerRestricted API Key authentication is down

    A critical outage around Restricted API Key authentication occurred Sunday evening at exactly 9:00 PM PST leading to over 1 hour of downtime. This outage did not result in any data loss but did render Restricted API Keys inoperable with our API endpoints for that period of time. This was an unacceptable amount of downtime and could have been completely prevented. We take your trust very seriously as Doppler is a critical path in your devops and productivity workflows. We have learned from this experience while fixing the root cause and adding checks to prevent this kind of outage in the future. Here is what happened: # Timeline **December 9th, 2019 - 9:00 PM \(PST\)** Our engineering team rolls out new authentication logic for our API endpoints. The update is designed to increase our defense in depth by adding additional layers of authentication around every layer in our core stack. **December 9th, 2019 - 9:40 PM \(PST\)** While testing the API endpoints to verify the new logic, we find that all requests utilizing a Restricted API Key are being rejected. **December 9th, 2019 - 10:00 PM \(PST\)** We identify the culprit bug in our codebase and start working on a patch. The bug is linked to how we handle Restricted API Keys. Doppler’s API offers 3 methods of authentication: Personal Keys, Restricted API Keys \(now called Service Tokens\), and CLI Tokens. Personal Keys and CLI Tokens are tied to a user identity, while Service Tokens are not. Our investigation finds that our authentication logic was requiring a user identity and did not gracefully handle a case where one would not be present. **December 9th, 2019 - 10:48 PM \(PST\)** The patch is released to production and the engineering team starts monitoring the fix. **December 9th, 2019 - 11:38 PM \(PST\)** After stress testing the patched authentication logic in production, we mark the incident resolved. # Moving Forward **End to End Testing \(e2e\)** From unit tests to e2e testing, we strive to test _every_ part of our stack. As it turns out, we had not added e2e tests for Service Tokens. This outage has led us to reassess our test coverage and focus on testing all remaining user flows. **StatusPage Accuracy** Doppler uses Pingdom to test for uptime and display those results on our status page. Pingdom was testing our health check endpoints which do not require authentication. This led Pingdom to not report our outage. We have now changed Pingdom to test our secrets endpoint, which requires multiple layers of authentication and is a better indicator of our API being serviceable to customers. **Announcing Maintenance Windows** Starting today, we are going to announce major maintenance schedules one week in advance. Our goal with maintenance is to _never_ have any downtime, but when downtime does strike, we want you to be prepared. Knowing a time window in advance can help your engineering teams navigate and prepare for the off chance appropriately.

    Resolved

  3. DopplerMigration Script Race Condition

    We had a migration script that failed to run before the rest of our services deployed. This caused the newly deployed services to crash because the database was missing a column in a critical table. Our deployment orchestrator quickly detected the issue and reverted to the previous build. Total downtime was approximately 2 minutes. We plan to mitigate future failures by requiring our migrations to complete before rolling out our remaining services.

    Resolved

  4. DopplerDowntime due to migration script

    Today at 8:38pm Pacific we experienced a critical outage that brought down all of our servers on both our primary and failover production environments. This outage did not result in any data loss but did disable all dashboard and API access. ## Why Did it Happen? At Doppler, we typically do batched rollouts to production at night \(lowest risk time\). To help prevent outages we use staged releases, where the migration scripts are run first, then our new code is released to our clusters. The problem we faced today with this approach is it creates an edge case when we are deleting a column in the database that is still being used in active code. In our latest rollout, we had a migration script that deleted a column in one of our tables. This column would not be used in the soon to be released code but was used by our active deployments. When the migration script ran, all our active deployments immediately crashed as the ORM \(object-relational mapper\) expected a column that did not exist anymore. ## Moving Forward **Migration Scripts** We are enacting a new policy where columns can not be deleted when there is an active deployment that relies on them. The new model will require that the column can only be deleted once 100% of our deployments are not using it. We expect this to come in the form of 2 rollouts, the first transitioning our deployments off the column with the second removing the column. **Hardening Our Deployments** One of the coolest parts of building tools for developers is that we get to dog food our own product. But it does come with its own struggles, such as Doppler relying on Doppler. Circular loops can be dangerous. To help prevent that, we create encrypted snapshots of our secrets in our Docker images during the build phase. This is done so that in the event an outage occurs, we can bring ourselves back up. The problem with the current approach is the timing of when those images are built. Currently, those images are built after the migration scripts have run. We will be shifting to a new model where all images are built before running migrations so that we have a guarantee we can access our secrets. The longest deployment step during a rollout is building our images. By building our images before running migrations, we get the added benefit of dramatically reducing the time between when the migration is run and our updated code is released. ## Wrapping Up We want to apologize to all of our customers for this unacceptable outage. We take your trust in our uptime very seriously. We have learned from this incident and will showcase it by continuing to do everything in our power to prevent future outages.

    Resolved

  5. DopplerDNS resolution failure

    Today at 2:15pm Pacific we experienced our second outage in the last two weeks. Sadly the timing is not great but it did give us the opportunity to re-evaluate our failure points as we continue to harden our infrastructure. This outage affected all of our endpoints, from our production and failover infrastructure to the documentation hub and status page but did not result in any data loss. ## What Happened? Doppler uses [Cloudflare](https://cloudflare.com) as our DNS provider which provides a suite of powerful features including DDOS protection, a CDN for assets, firewall rules, edge workers, and plenty of others. They are one of the most popular and trusted DNS providers, which supports nearly 20% of all internet traffic. Today they went down, which brought down a portion of the global internet with them. Cloudflare recommends using their DNS proxy so you can benefit from their suite of features. As we were reminded today, using their proxy changes the landscape of the default protections DNS provides, which results in a nonobvious cost. DNS by its very nature is decentralized, which creates a layer of resilience against being a single point of failure. But this assumption of protection breaks down when you use a proxy at the DNS layer, as now you have a new single point of failure. Today we all paid that nonobvious cost. ## Moving Forward **Hardening Our DNS Reliability** Internally we are tracking the best path forward towards hardening our DNS’s reliability. This can come in a couple of different forms, such as disabling proxy mode for our DNS records. This would remove our DNS layer as a single point of failure but comes at the cost of losing DDOS protection, our records not being masked, and also some other behind the scenes magic. Another possible option would be to add an additional DNS provider \(that supports DDOS protection\) to our stack. Then in the case one goes down, our traffic will automatically failover the other. This would add a fair amount of complexity to our stack. Sadly all solutions thought of so far have tradeoffs and could have nonobvious consequences. We deeply care about finding the right answer, not the fastest to implement. As we continue to explore and implement solutions, we expect to write about our findings and decisions on our [engineering blog](https://doppler.com/blog). **Customer Observability** Being transparent is core to the DNA of the company and we strive to provide our customer's observability during outages in real-time. We do this through our [@DopplerHelp](https://twitter.com/DopplerHelp) Twitter and [status page](https://status.doppler.com). Being that our status page’s DNS is hosted by Cloudflare, it was also affected by the outage. To prevent this in the future, we are moving our status page’s DNS to another provider and will use a new dedicated domain. This domain is still being configured and will be announced soon. **Doppler CLI** The Doppler CLI has a nifty command called `doppler run` which downloads your secrets from our API and then injects them into your application. After each successful run, we automatically create and store an encrypted snapshot of the secrets for you. On the off chance the CLI is unable to connect to our API, we smartly fallback to this encrypted snapshot after 5 retries. During the outage, our `doppler run` users were unaffected as they had an existing snapshot to fallback to. One area we found that could use a little love is in showcasing of a retry event. If request hangs it can create a visible delay to the user. In the next release, the Doppler CLI will print a message stating a retry event is happening so you always stay informed. ## Wrapping Up Providing a seamless experience that provides near perfect uptime is an incredibly difficult task that requires deep thought about every layer in the stack. Today we are reminded that our DNS is a single point of failure and that even with the most trusted of services, like Cloudflare, can bring us down if we don’t have multiple layers of redundancies. As we continue to harden our infrastructure, we plan to share our learnings with you through our [engineering blog](https://doppler.com/blog).

    Resolved

Jul 23, 2020

Jul 18, 2020

  1. UserVoiceDNS Failure Causing Site Availability Issues

    Cloudflare believes a bad router announcement caused massive backend failures on their system. This caused DNS to fail for multiple regions (explicitly North America). The issue has been fully resolved and tested.

    Resolved

Jul 17, 2020

Jul 10, 2020

Jul 8, 2020

Jul 3, 2020

Jun 11, 2020

Jun 9, 2020

Jun 8, 2020

  1. GradleAn issue with publishing keys

    Plugin publishing has been restored. New user accounts that were created in the last 12 hours may need to regenerate the publishing keys for their profile. We're sorry for the inconvenience.

    Resolved

Jun 3, 2020

  1. CourseraCoursera.org unavailable

    Coursera.org is now available. Thank you for your patience while we resolved the issue, and we apologize for the inconvenience.

    Resolved

Jun 1, 2020

  1. Scout APMIngestion Issue

    On 2020/05/31 we experienced a short network outage that prevented our zookeeper and kafka nodes from reaching each other. When connectivity was restored, there was a problem with stale zookeeper data which prevented the kafka brokers from initiating a proper leader election for topic partitions. This also prevented kafka producers from being able to produce to a majority of partitions. A manual leader election was attempted, but failed to correct the issue. We began a rolling restart of our entire kafka cluster, which ultimately resolved the issue. Later versions of Kafka have better handling around this particular failure, and we anticipate moving to a recent version to prevent entering this failure mode again.

    Resolved

May 30, 2020

May 5, 2020

  1. CourseraSitewide outage

    This issue has been resolved, and all coursera.org pages should load normally.

    Resolved

Apr 29, 2020

Apr 28, 2020

Older incidents