Cases12 cases

7 to 11 of 12

1. IL-0216 Wikimedia on-wiki search was slow or rejected after a datacenter switchover overloaded one Elasticsearch node in codfw

   Wikimedia Foundation2023

   Symptoms

   Right after the primary datacenter was switched from EQIAD to CODFW at 14:00 UTC on 20 September 2023, on-wiki search on all Wikimedia wikis became delayed or unavailable for a subset of users. At 14:08 the alert "CirrusSearch codfw 95th percentile latency - more_like" fired and reached IRC and the Search Platform email list as "PROBLEM alert - graphite1005/CirrusSearch codfw 95th percentile latency - more_like is CRITICAL".

   High latencyHigh error rateLoad imbalanceMisrouted requests

   Trigger

   At 14:00 UTC the primary datacenter was switched from EQIAD to CODFW. more_like queries that were expected to go to eqiad, the non-primary datacenter, went to codfw instead, apparently because RESTBase sent them as POST rather than GET.

   Root cause

   Software defect

   Component

   Elasticsearch or OpenSearchIn-memory cache

   Contributing factor

   Unverified assumptionDependence on a warm cacheNo runbook

   Detection

   Automated monitoring caught the problem, but not quickly enough to prevent user-facing impact: the Grafana alert "CirrusSearch codfw 95th percentile latency - more_like" went to IRC and the Search Platform email list at 14:08. No one was paged; SREs were already watching closely because of the datacenter switchover. At 14:14 Hashar pinged #wikimedia-search and engineers began troubleshooting.

   Remediation

   • Banned the overloaded node elastic2044 from the Elasticsearch cluster

   Action items

   • Work out which Elasticsearch steps are needed before a datacenter switchover
   • Update the Elasticsearch section of the datacenter switchover documentation
   • Review the suggested switchover improvements in T285347 and decide whether to implement them
   • Make the RESTBase /related/{article} endpoint send GET instead of POST so the CirrusSearch query cache stays warm in both datacenters
   • Review the shard balance of busy wikis, especially enwiki_content

   Lessons

   Monitoring caught the problem, but not quickly enough to prevent user-facing impact. The heightened attention during the switchover meant the impact was recognised immediately, and banning a single node was quick and easy. The mitigation was not guided by an existing runbook, and the shards on the overloaded host were not checked before it was banned.

   View case
2. IL-0459 MediaWiki on Kubernetes terminated TLS with a wrong certificate after a deployment, and 1% of requests failed

   Wikimedia Foundation2023

   Symptoms

   On 9 August 2023, right after a MediaWiki configuration deployment at 11:41 UTC, the MediaWiki deployments on Kubernetes terminated TLS with a wrong certificate, and about 1% of Wikimedia web and API traffic, around 120 requests per second, received errors for about 30 minutes. The issue was detected by a page for failed service probes on mw-api-ext and mw-api-int.

   High error rateTLS certificate errorsControl plane lag

   Trigger

   At 09:37 a change to the global Kubernetes defaults was merged that switched supported charts to cert-manager certificates. The next MediaWiki on Kubernetes deployment, a configuration change at 11:41, picked it up.

   Root cause

   Configuration change

   Component

   cert-managerHelm

   Contributing factor

   Change with a delayed effectNo configuration tests

   Detection

   A page from the ProbeDown alerts for mw-api-ext:4447 and mw-api-int:4446 at 11:47; a KubernetesAPILatency alert had fired at 11:46.

   Remediation

   • Identified the issue and prepared a patch, https://gerrit.wikimedia.org/r/947331
   • Redeployed the MediaWiki deployments on Kubernetes

   Action items

   • Review whether deployment procedures and testing should be strengthened, for example against surprising changes on the next deployment and with canary deployments for Kubernetes
   • Look at metrics that become unavailable or unhealthy during deployments, or at reducing the impact of deployments

   Lessons

   The cause was found quickly because the person who made the breaking change was responding. The change should have been caught by CI, but the repository it was made in (hieradata) does not run the CI of the deployment charts. Only 1% of traffic going to Kubernetes kept the impact small, and the low error volume made it slower to link the errors to Kubernetes.

   View case
3. IL-0267 Wikimedia's eqsin cache site went unreachable when both of its transport links to the core data centers were down

   Wikimedia Foundation2023

   Symptoms

   On 10 January 2023 at 16:37 UTC Icinga reported "BGP status on cr3-eqsin is CRITICAL: BGP CRITICAL - AS1299/IPv6: Active - Telia, AS1299/IPv4: Active - Telia", and from 16:45 the hosts in Wikimedia's eqsin site went DOWN with "PING CRITICAL - Packet loss = 100%". Users normally directed to eqsin, mostly in the APAC region, could only read Wikipedia pages already cached in eqsin.

   Network unreachable

   Trigger

   Planned maintenance by the transport provider took down eqsin's second transport link while the first was still down from a long fiber cut.

   Root cause

   Dependency failure

   Component

   Backbone network (WAN)BGP routingCDN and edge network

   Contributing factor

   Single point of failureIgnored early signal

   Detection

   Automated monitoring: Icinga BGP status alerts for cr3-eqsin at 16:37 and cr2-eqsin at 16:44, followed by a flood of host-down alerts and three pages at the same time.

   Remediation

   • Depooled eqsin from DNS
   • Repooled eqsin

   Action items

   • Create a backup GRE tunnel for eqsin (T327265)
   • Ensure the long-down transport link comes back up properly once it is fixed (T322529)
   • Automatically parse maintenance notifications and alert on conflicting maintenance (T230835)

   Lessons

   The site was depooled quickly, and "depool first, investigate later" was the right attitude. A flood of host-down alerts usually means a network problem. eqsin stayed too long on a single working transport link, SREs did not catch that the downtime of the two transport links overlapped, and the provider's planned maintenance was not recorded in a task or reminder.

   View case
4. IL-0266 Eqiad routers looped traffic between each other after a BGP MED change and could not reach other sites' public IPs

   Wikimedia Foundation2021

   Symptoms

   On 23 November 2021 at 09:39, two minutes after a change on the eqiad core routers, alerts fired because Icinga in eqiad could not reach the public text-lb address in eqsin (Singapore). For about 12 minutes eqiad could not reach hosts in other data centers, such as the cache PoPs, through their public IP addresses.

   Network unreachable

   Trigger

   At about 09:37 a change on cr1-eqiad and cr2-eqiad, made as part of task T295672 to influence BGP route selection, removed a BGP setting that sets the BGP "MED" attribute to the OSPF cost of reaching the BGP route's next hop.

   Root cause

   Configuration change

   Component

   BGP routing

   Contributing factor

   Ineffective rollback

   Detection

   Icinga health checks against edge caching clusters alerted at 09:39; the first alert was Icinga in eqiad failing to reach the public text-lb address in eqsin.

   Remediation

   • Reverted the configuration change on both eqiad routers
   • Forced a reset of the BGP session between the two eqiad routers

   Action items

   • Updated T295672 with further detail on the incident and a proposal to adjust the BGP sessions to address the "second IXP port" issue and prevent this from happening again

   View case
5. IL-0323 A RAID battery backup unit failure crashed db1075, the MariaDB primary of section s3, and every s3 wiki went read-only

   Wikimedia Foundation2019

   Symptoms

   At 18:40 UTC db1075, the primary database of section s3, stopped answering, and all s3 wikis went read-only because no writes could reach it. Reads kept working from the replicas. At 18:42:45 Icinga reported "PROBLEM - Host db1075 is DOWN: PING CRITICAL - Packet loss = 100%".

   Writes rejectedReplication lagHigh latency

   Trigger

   The RAID battery backup unit (BBU) of db1075, the s3 primary database host, failed and the whole host crashed.

   Root cause

   Hardware failure

   Component

   DatabaseDatabase replication

   Contributing factor

   Pages not deliveredHarmful manual action

   Detection

   Icinga reported db1075 HOST DOWN on IRC at 18:42:45, but that alert does not page; replica Slave IO alerts followed, users reported problems on #wikimedia-operations, and the first SMS page went out at 18:56. Volunteers and staff noticed the failure before the alerts did.

   Remediation

   • Power-cycled db1075 from the iDRAC after finding the BBU problem
   • Let MySQL finish InnoDB recovery on db1075
   • Removed read_only=ON from db1075 by hand
   • Ran puppet by hand a second time so pt-heartbeat started

   Action items

   • Implement or refactor a script that moves replicas when the primary is unavailable
   • Fix the MediaWiki heartbeat model so pt-heartbeat does not use a super-user, avoids a single point of failure and follows the real primary without depending on puppet
   • Decide what to do with the batch of hosts that already had BBU issues
   • Buy a new BBU for db1075
   • Remove db1075 from the primary role
   • Address MediaWiki spam while the database is read-only or the primary is unavailable
   • Make primary database hosts page on HOST DOWN
   • Track hardware errors better in Netbox

   Lessons

   A BBU failure should not crash a whole host, yet it had happened before on HP hosts. HOST DOWN on a primary database should page instead of only alerting on IRC. Starting pt-heartbeat through puppet races with MySQL startup, and the unnoticed failure kept lag reported and made the outage 8 minutes longer.

   View case

By root cause7