Vault Error Checking In Code Updates

davenovak · Post by **davenovak** » Thu May 03, 2007 4:52 pm

I am using Vault 3.5.1 client and server.

Twice in the past week I've see an issue where a programmer tries to check in a code update and they get a Vault error. Here's the stack dump from the server log:

Code: Select all

----5/3/2007 6:11:04 PM      Kharmon--192.168.104.159(192.168.104.159)--SSL Disabled	BeginTx: Critical Error! Error in the application.
   at VaultServiceBase.VaultFolder.DeleteFolderEntryByName(HybridDictionary htParentObjects, String strSearchName, Int32 nObjType, Int32 nFSOHashCode)
   at VaultServiceBase.VaultRepository.CreateFolderReferenceCopy(Int64 nRevID, HybridDictionary hdNewRefParents, VaultFolders vfOwners, VaultFolder vfOld)
   at VaultServiceBase.VaultRepository.GetReferenceCopiesByPath(Int64 nRevID, HybridDictionary hdNewRefParents, String strAbsolutePath, Boolean bRefCopyLastObjectWithAllFolders, VaultFSObjects& fsoList, Int32& nPinCount)
   at VaultServiceAPILib.VaultTransactionContainer.GetReferenceCopiesByPath(String strAbsolutePath, VaultFSObjects& fsoList, Int32& nPinCount)
   at VaultServiceAPILib.VaultTransaction.PreCheckCheckIn(Int32 nCurrStatCode, VaultRequestCheckIn vrci, String strFileToken, VaultResponseCheckIn vRespCheckIn, VaultTransactionContainer txContainer)
   at VaultServiceAPILib.VaultServiceAPI.BeginTx(HttpApplicationState theApp, VaultLoginInfo vliLoggedInUser, Int32 nTxUserID, Int32 nRepID, VaultDateTime dtBeginTx, String strComment, VaultRequestItem[]& requests, String& strTxID, VaultIntTx& vit)
   at VaultService.VaultService.DoBeginTx(HttpApplicationState has, HttpSessionState hss, VaultLoginInfo vli, String strCallerLogEvt, Int32 nTxUserID, Int32 nRepID, VaultDateTime dtBeginTx, String strComment, VaultRequestItem[]& requests, String& strTxID)

I don't know what might be going wrong here, but after clearing the local Vault client cache on his machine, the problem was resolved (for the second time now in a week).

Is anybody looking into what the root cause of such problems could be? A real fix here would be nice.

In one of my posts a few months ago (http://support.sourcegear.com/viewtopic.php?p=30371) there really was no answer for either 1) why the cache gets corrupt or 2) why clearing the caches fixes it. In yet another post of mine (http://support.sourcegear.com/viewtopic.php?=&p=30128) this same problematic method DeleteFolderEntryByName is listed as the top function in the stack.

I sincerely hope that someone is taking a good hard look at the code in DeleteFolderEntryByName as it is clearly not as robust as it needs to be.

Beth · Post by **Beth** » Fri May 04, 2007 8:55 am

Could you perform an iisreset at the server command prompt instead of clearing the client cache next time?

So far what we've been able to determine is that this could be due to the way file and folder sharing is being used. If you would like to dig further, could you send an email to support at sourcegear.com (attn: Beth) with a link to this post?

davenovak · Post by **davenovak** » Fri May 04, 2007 9:03 am

E-Mail on its way