Closed HakanL closed 11 months ago
I get this over and over:
2023-12-06T15:35:48.351999035Z [15:35:48 INF] Seq 2023.4.10219 running on OS Linux 5.15.131.1-2.cm2 #1 SMP Sun Sep 24 03:38:45 UTC 2023
2023-12-06T15:35:50.000149013Z [15:35:49 INF] Opening event store at /data/Stream/stream.flare
2023-12-06T15:35:50.290555012Z [15:35:50 INF] Ingestion enabled
2023-12-06T15:35:50.349534621Z [15:35:50 INF] Opening metastore /data/Documents/metastore.flare
2023-12-06T15:35:50.626346179Z [15:35:50 INF] Closing event store
2023-12-06T15:35:50.660314530Z Application startup exception: Autofac.Core.DependencyResolutionException: An exception was thrown while activating λ:Seq.Cluster.IClusterNode -> λ:Seq.Server.StorageSubsystem.
2023-12-06T15:35:50.660348230Z ---> Flare.Ffi.Result.FlareException: Flare native storage failed (IOError), error evaluating the next document to read.
2023-12-06T15:35:50.660373130Z caused by: snappy: error snappy reading buffered frame header
2023-12-06T15:35:50.660378730Z caused by: snappy: corrupt input (unsupported chunk length: 107)
2023-12-06T15:35:50.660401330Z at Flare.Ffi.Result.FlareResult.EnsureSuccess()
FWIW I'm using Azure Files, but I was under the impression that it's supported now without corruption, it's set to Premium LRS. This data in this SEQ instance isn't super-critical, and I'm happy to share files from it if it can help figuring out the issue. But I'd prefer not to wipe it if possible.
Update: I was able to delete the files in the Documents folder and restore the last auto-backup, so I'm back up and running. But it would be interesting to know what could cause this, as I mentioned I thought using Azure Files would be supported? Or is there an issue in how I host it (App Service)?
Hi @HakanL 👋
Glad you got it all sorted. This looks like it was a data corruption. Unfortunately, finding properly reliable storage options in Azure is hard. Premium Files still occasionally corrupts, but I think it is as good as it gets outside of VMs or AKS. For more critical workloads, you can move your document storage to SQL Azure. We’ve got some docs on that here: https://docs.datalust.co/docs/using-sql-server-as-a-metastore
@KodrAus Is the metastore more "sensitive" than the stream storage? Or are both susceptible to the corruption?
Both are susceptible to corruption but the stream storage is definitely more corruption resistant. Most of your event stream is chunked in immutable files, and if we encounter a corruption in recently ingested data then you might lose up to a few minutes worth of logs, which we just log as a metric typically.
Describe the bug After a clean shut down my instance fails to start. I'm getting this error in the docker/seq log:
Environment (please complete the following information):
Additional context Here's a part of the log file: