Hi everyone,
We’ve just released ServiceControl 6.22.3.
Dependent CVEs
- #6110 System.Formats.Cbor denial-of-service vulnerability (CVE-2026-118747)
Security updates
Fixed bugs
- #6124 Maintenance mode fails to start when a locked index has a changed definition
- #6122 Event log queries create the RavenDB auto index Auto/EventLogItems/ByRaisedAt in the error database
- #6123 Throughput queries create RavenDB auto indexes in the throughput database
ServiceControl 6.22.3 includes the .NET 10.0.15 runtime. This runtime contains the November 2026 ASP.NET Core security updates. ServiceControl 6.22.3 also updates the System.Formats.Cbor package, which the Amazon SQS transport uses, to version 10.0.15. This version resolves CVE-2026-118747, a denial-of-service vulnerability in System.Formats.Cbor.
The other three bugs apply to error and audit instances that use RavenDB. Monitoring instances do not use RavenDB and are not affected.
How to know if you are affected
- #6104 applies to all users of ServiceControl 6.22.2 and earlier.
- #6110 concerns a package that error instances use with the Amazon SQS transport. ServiceControl 6.22.3 updates the package as a precaution.
- You are affected by #6124 if you upgrade a Windows deployment of an error or audit instance to 6.22.2, and the upgrade changes the definition of an index that you locked in RavenDB Studio.
- You are affected by #6122 if you use an error instance and ServicePulse shows the event log. This applies to most error instances. The symptoms are visible from 6.20.0.
- You are affected by #6123 if you use an error instance from version 5.4.0 or later that collects throughput data for usage reports. The symptoms are visible from 6.20.0.
Symptoms
- For #6124, the instance stops when you start it in maintenance mode. The start-up error names the locked index. The embedded RavenDB database stops with the instance, so you cannot open RavenDB Studio to unlock the index.
- For #6122 and #6123, the List of Indexes in RavenDB Studio shows indexes with names that start with
Auto/. For #6122, the index isAuto/EventLogItems/ByRaisedAtin the error database. For #6123, the indexes are onEndpointDocumentsin the throughput database. The search engine migration procedure cannot change these auto indexes. Thus, after you migrate all indexes to Lucene, the instance log continues to show the warningThe following RavenDB index(es) use the Corax search engineat start-up. TheError Database Search Enginecustom check also continues to fail in ServicePulse.
When to upgrade
You should upgrade immediately. This release contains security updates.
In 6.22.3, maintenance mode does not perform database setup. It does not create databases or deploy indexes. If a locked index stops the start-up after the upgrade, follow these steps:
- Start the instance in maintenance mode.
- Unlock the index in RavenDB Studio.
- Start the instance in the usual mode.
At start-up, error instances on 6.22.3 add two static indexes:
EventLogItems/ByRaisedAtin the error database.EndpointIndexin the throughput database.
These indexes are small. RavenDB does not rebuild existing indexes.
ServiceControl does not delete the existing auto indexes. Until they are deleted, the start-up warning and the Error Database Search Engine custom check continue to report them. RavenDB marks an auto index as idle after 30 minutes without queries, and deletes it 72 hours later. To clear the warning sooner, manually delete the auto indexes in RavenDB Studio.
ServiceControl also deletes the unused EndpointIndex from the error database at start-up. If this index is locked, ServiceControl unlocks it before it deletes it.
Where to get it
You can download the new version of ServiceControl from our website or Docker Hub.
Please read our support policy for more details.
With thanks,
The team in Particular