Bugcheck Type: ASSERTFAIL, System ASSERT failure detected

[Toine_1]

Hi,

I got a crash dump on a Integrity server.
I run VMS8.3-1H1.
What does this mean "System ASSERT failure detected?"

Has anyone also had this problem?

Crash Time: 12-MAR-2010 08:45:06.58
Bugcheck Type: ASSERTFAIL, System ASSERT failure detected
Node: NVR (Cluster)
CPU Type: HP BL860c (1.67GHz/9.0MB)
VMS Version: V8.3-1H1
Current Process: TAX_ARCHDSO
Current Image: DSA101:[PRSTODISK.][STO.EXE]ARCHDSO.EXE;10
Failing PC: FFFFFFFF.80D40680 SYS$XFCACHE+1C280
Failing PS: 00000000.00000800
Module: SYS$XFCACHE (Link Date/Time: 14-NOV-2008 08:05:21.73)
Offset: 0001C280


/Toine

[Volker Halle · Accepted solution]

Toine,

this is a software problem in the XFC code (eXtended File Cache).

Could you attach the full CLUE file (CLUE$COLLECT:CLUE$NVR_12MAR2010_0845.LIS file as an attachment ?

The most recent XFC patch is VMS831H1I_XFC-V0200 (released 06-NOV-2009), you don't have this patch installed. Please consider to upgrade to the most recent UPDATE patch for OpenVMS I64 V8.3-1H1.

Volker.

[Volker Halle]

Toine,

typo correction, the CLUE file should be:

CLUE$COLLECT:CLUE$NVR_12032010_0845.LIS

VMS831H1I_UPDATE-V0700 contains VMS831H1I_XFC-V0200, which is the most recent available patch for XFC.

An ASSERTFAIL crash is not included in the list of 'problems fixed' for that ECO, but - even if you escalate this crash to HP now - you'll probably be asked to 'install the most recent patch'.

Volker.

[Toine_1]

Volker,

I have uploaded the clue fie.

/Toine

[Volker Halle]

Toine,

no obvious problems in the CLUE file. Just a software crash in XFC.

Please open a call with HP and be prepared to be asked for installing the most recent patches.

Volker.

[P Muralidhar Kini]

>> What does this mean "System ASSERT failure detected?"

XFC does a lot of consistency checking in its code and the moment it detects any inconsistency it initiates a inline bugcheck "ASSERTFAIL, System ASSERT failure detected" to crash the system.


>> I have uploaded the clue fie.
I had a quick look at the clue file and the following are my findings -

When XFC was trying to DEQ a lock associated with a file (most probably the lock here is associated with a file, however it can also be a lock associated with a volume, need the crash dump to know exactly whether it's a file or volume) it got SS$_IVLOCKID error. The error handling code for SS$_IVLOCKID error needs a XFC context block to proceed. However in certain scenarios, XFC does
not pass a context block to this routine. It's in these scenarios that XFC crashes the system when it gets a SS$_IVLOCKID error.

We need the full crash dump in order to get further clues as to why XFC got a SS$_IVLOCKID error while it was trying to DEQ a lock.

[Toine_1]

Hello,

I will send a full crash dump today to HP.

/Toine