Index errors after upgrade to 3.5

[Matthias De Ridder]

We have upgraded RavenDB from version 30000 to version 35183. Since then, three indexes are throwing a lot of errors:

Could not map file D:\InfohosServices\PROD\Financieel\Tarfac\Databases\Databases\tarfac.027\Indexes\86\_6zy.fdt

Due to the error count, the indexes become disabled. In the log, there is an 0x80004005 error code. 

We tested the upgrade several times and never encountered this error. We have no idea what to check right now. Any suggestions where to look for the problem?

[Matthias De Ridder]

This is an example of one of the errors in the log file:

2017-01-17 16:20:47.7510;Raven.Database.Indexing.Index.Indexing;Warn;;Failed to reduce index: PaymentPlan/BillingStatus (id: 80) 7 single step keys and 0 multi step keys. Skipping this batch (it won't be reduced);"System.ComponentModel.Win32Exception (0x80004005): Could not map file D:\InfohosServices\PROD\Financieel\Tarfac\Databases\Databases\tarfac.027\Indexes\80\_19h0.fdt

   at Raven.Database.Indexing.LuceneCodecDirectory.CodecIndexInput..ctor(FileInfo file, Func`2 applyCodecs)

   at Raven.Database.Indexing.LuceneCodecDirectory.OpenInputInner(String name)

   at Raven.Database.Indexing.LuceneCodecDirectory.OpenInput(String name, Int32 bufferSize)

   at Lucene.Net.Store.FSDirectory.OpenInput(String name)

   at Raven.Database.Indexing.LuceneCodecDirectory.FileLength(String name)

   at Lucene.Net.Index.SegmentInfo.SizeInBytes()

   at Lucene.Net.Index.LogMergePolicy.SizeBytes(SegmentInfo info)

   at Lucene.Net.Index.LogByteSizeMergePolicy.Size(SegmentInfo info)

   at Lucene.Net.Index.LogMergePolicy.FindMerges(SegmentInfos infos)

   at Lucene.Net.Index.IndexWriter.UpdatePendingMerges(Int32 maxNumSegmentsOptimize, Boolean optimize)

   at Lucene.Net.Index.IndexWriter.GetReader(Int32 termInfosIndexDivisor)

   at Lucene.Net.Index.IndexWriter.GetReader()

   at Raven.Database.Indexing.Index.RecreateSearcher()

   at Raven.Database.Indexing.Index.Write(Func`4 action, List`1 writePerformanceStats)

   at Raven.Database.Indexing.MapReduceIndex.ReduceDocuments.ExecuteReduction()

   at Raven.Database.Indexing.IndexStorage.Reduce(Int32 index, AbstractViewGenerator viewGenerator, IEnumerable`1 mappedResults, Int32 level, WorkContext context, IStorageActionsAccessor actions, HashSet`1 reduceKeys, Int32 inputCount)

   at Raven.Database.Indexing.ReducingExecuter.SingleStepReduce(IndexToWorkOn index, List`1 keysToReduce, AbstractViewGenerator viewGenerator, ConcurrentSet`1 itemsToDelete, Boolean skipIncreasingBatchSize, CancellationToken token)

   at Raven.Database.Indexing.ReducingExecuter.HandleReduceForIndex(IndexToWorkOn indexToWorkOn, Boolean skipIncreasingBatchSize, CancellationToken token)"

[Oren Eini (Ayende Rahien)]

Are you running in 32 bits mode?

Hibernating Rhinos Ltd  

Oren Eini l CEO l Mobile: + 972-52-548-6969

Office: +972-4-622-7811 l Fax: +972-153-4-622-7811

 

--
You received this message because you are subscribed to the Google Groups "RavenDB - 2nd generation document database" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ravendb+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Matthias De Ridder]

How can we check this? RavenDB is hosted in IIS with the setting enable 32-Bit Applications on true.

Op dinsdag 17 januari 2017 17:19:00 UTC+1 schreef Oren Eini:

To unsubscribe from this group and stop receiving emails from it, send an email to ravendb+u...@googlegroups.com.

[Oren Eini (Ayende Rahien)]

Disable that, and it will resolve the issue.

To unsubscribe from this group and stop receiving emails from it, send an email to ravendb+unsubscribe@googlegroups.com.

[Matthias De Ridder]

We found the cause: we installed the new RavenDB version with the installer while selecting the existing application pool. When you choose this option, the setting enable 32-Bit Applications is set to true by the installer.

Op dinsdag 17 januari 2017 17:31:54 UTC+1 schreef Oren Eini:

[Matthias De Ridder]

Why does it go wrong when it runs in 32 bit?

Op dinsdag 17 januari 2017 18:20:43 UTC+1 schreef Matthias De Ridder: