Cases45 cases

16 to 20 of 45

1. IL-0024 Expensive cache-backed API calls and uncancelled database requests overloaded MongoDB and took the Hugging Face Hub down three times

   Hugging Face2024

   Symptoms

   At 08:45 CET on 22 April 2024 the Hugging Face Hub became unstable, and by 09:30 it was unresponsive. For most users the website was slow or could not be reached at all.

   Service outageHigh latencyDatabase overload

   Trigger

   Bursts of requests to revision endpoints that read heavily from the *_info_cache collections: a call to https://huggingface.co/api/datasets/lighteval/mmlu/revision/main rose sharply right before the first crash, and a call to https://huggingface.co/api/models/sentence-transformers/all-mpnet-base-v2/revision/main, a model with 300+ associated spaces, lined up with the crash at 21:30 on 23 April.

   Root cause

   Software defect

   Component

   Application databaseModel hubCluster autoscaler

   Contributing factor

   No headroomRetry stormOverlapping failures

   Detection

   Instability started at 08:45 on 22 April and the Hub was unresponsive by 09:30; the team followed MongoDB Atlas metrics and a MongoDB support case. The root cause was found days later from a new graph of requests per URL.

   Remediation

   • Added a firewall rule blocking all requests with a Java/* user agent
   • Disabled the quicksearch feature on the Hub
   • Scaled the MongoDB Atlas cluster by hand from tier 2 to tier 3
   • Switched the Hub to maintenance mode with a 503 page through AWS WAF and blocked incoming traffic
   • Scaled all Hub pods down to 0 and brought them back
   • Set an aggressive rate limit
   • Added timeouts to requests and reworked database queries that were slow or returned large bodies

   Action items

   • Add timeouts to requests based on their usual completion time
   • Optimise requests that took too long or returned a large body, and rework poorly designed database queries
   • Remove old unused indexes from the database
   • Build new dashboards and monitoring metrics
   • Move the *_info_cache collections to Redis
   • Make it possible to abort any database request when the client aborts theirs

   Lessons

   Finding the root cause took two to three days: the existing charts did not show it, and a per-URL request breakdown did. Not cancelling database work when a client goes away turns a strained database into a thundering herd. A close inspection of the database exposed architectural flaws and bottlenecks, such as a collection with large unused indexes, and taught the team which metrics need closer watching to anticipate crashes. Unrelated failures happening at the same time added to the stress and made restoring service harder.

   View case
2. IL-0161 A faulty earlier API update made new LKE cluster provisioning in Toronto time out for three days, leaving clusters with no node pools or the wrong number of nodes

   Linode2024

   Symptoms

   At about 14:05 UTC on 22 April 2024, Linode's team noticed that new LKE clusters were failing to provision in the Toronto data center.

   Cluster or node provisioning stallIntermittent timeouts

   Trigger

   A previous update to the API did not work as intended and caused timeouts when new LKE clusters were provisioned.

   Root cause

   Software defect

   Component

   Node pool

   Contributing factor

   –

   Detection

   Linode's team noticed new LKE clusters failing to provision in the Toronto data center at about 14:05 UTC on 22 April 2024.

   Remediation

   • disabled the affected section of Toronto infrastructure from provisioning new LKE clusters
   • implemented a fix for the faulty API update
   • put the disabled infrastructure back into use after the fix

   Lessons

   The fault was in an earlier API change, not in the infrastructure that showed it, but most failures concentrated on one backend resource in one data center. Taking that resource out of provisioning limited the impact while the root cause was found, which took three days.

   View case
3. IL-0173 cert-manager's preferredChain logic picked the Let's Encrypt chain through the expired DST Root CA X3, so clients built on an old BoringSSL could not connect to a development server

   Devsisters2024

   Symptoms

   On 13 February 2024, when Devsisters' infrastructure cell came back from the Lunar New Year holiday, it received a request to look into a development server that could not be reached. No one had made a manual change over the holiday.

   TLS certificate errors

   Trigger

   On 8 February 2024, in the first phase of retiring the DST Root CA X3 cross-sign, Let's Encrypt made the ISRG Root X1 chain its default chain and began offering the DST Root CA X3 chain as the alternate.

   Root cause

   Software defect

   Component

   cert-manager

   Contributing factor

   –

   Detection

   On 13 February, after the Lunar New Year holiday, a request came in to the infrastructure cell that the development server could not be reached; while inspecting the certificate chain the server presented, the team found it serving the chain through the expired DST Root CA X3.

   Remediation

   • submitted a pull request fixing cert-manager's preferred chain selection, which was merged

   Lessons

   A latent bug can stay hidden until an upstream provider changes a default. cert-manager's preferredChain matching gave the intended result while DST Root CA X3 was Let's Encrypt's default chain and the wrong one after the default changed. An expired root at the top of a served chain is harmless to most TLS clients, which fall back to the trusted ISRG Root X1 below it, but fatal to an implementation such as an old BoringSSL that checks the trust anchor's validity and insists on verifying the topmost certificate. ACME clients choose the preferred chain with different heuristics, so the same preferredChain value can behave differently between Certbot and cert-manager.

   View case
4. IL-0065 Any pod could take over the cluster by rewriting Argo CD's unauthenticated Redis cache (CVE-2024-31989)

   Cycode2024

   Symptoms

   This is a vulnerability report, not an outage, so nothing failed visibly. What an engineer can check is whether the Argo CD Redis server accepts connections without a password from a pod in another namespace, on port 6379 at the address kube-dns returns. In the default installation it did.

   Unauthorised access or a foreign workload

   Trigger

   Cycode Labs installed Argo CD with the default configuration on a new Amazon EKS cluster and connected to its Redis server from a low-privilege pod in another namespace, simulating a compromised pod.

   Root cause

   Software defect

   Component

   Argo CD

   Contributing factor

   Unreviewed defaultsExcessive blast radiusCredential in configuration

   Detection

   Found by Cycode Labs security research, not by an operator.

   Remediation

   • Argo CD released 2.11.1, 2.10.10, 2.9.15 and 2.8.19 with an initialized password for the Redis server

   Lessons

   The hash that checks the cached manifests is computed without a secret key, so anyone who can write to Redis can recompute it and make Argo CD deploy an altered manifest. Cycode recommends upgrading to the fixed versions, checking that the argocd-redis-network-policy NetworkPolicy exists and that a CNI enforces it, using controller-based secret management instead of injecting secrets through Argo CD plugins so secrets never reach Redis, and running Argo CD on a dedicated cluster.

   View case
5. IL-0123 A response buffer allocated anew in a loop made OpenAI's routing layer nodes hit their memory limits and fail readiness checks on a morning with more completions than any day before, causing 502 and 503 errors across all models

   OpenAI2023

   Symptoms

   Between 5:42 AM and 7:16 AM PT on 8 November 2023 a large share of requests to OpenAI failed with 502 or 503 errors. Every model and API endpoint saw significant failures.

   High error rateService outageProbe failures

   Trigger

   A morning with dramatically more completions than any previous day pushed the routing layer nodes, which already had a chronic memory problem, past their memory limits.

   Root cause

   Software defect

   Component

   Model request routingContainer memory limit

   Contributing factor

   No alert on limitNo headroom

   Remediation

   • limited incoming traffic
   • redeployed the routing service en masse
   • ramped traffic back up slowly

   Action items

   • Pre-allocate the responseBuffer and reuse it instead of allocating a new one in a loop (3X improvement in memory and CPU usage)
   • Adjust the configured memory limits to an appropriate level so the service has significant headroom
   • Add rate limit controls to shed traffic more gracefully
   • Increase the service's capacity
   • Change alerting to catch the underlying memory behaviour before it escalates into service issues
   • Configure autoscaling for the service, which was not possible before because added capacity could hurt an upstream service

   View case

By root cause1