Trace files has error 2750

[Former Member]

Hi,

Getting error in trace files :

CLIENT_TRACE TRexAPpiAdmin.cpp(168330:show cannot determine index server for index xi_msg:af.SID.hostname

Error:Document status was not returned from the server.

SERVER_TRACE TRexApiadmin.cpp(16978):Index ID:xi_msg:af.sid.hostname, NS_SEARCH_DELTA lock failed in showIndex rc=2750.

Regards.

[former_member217429]

Hi Ambarish,

the error means usually that the index is corrupted. What's the TREX version you use?
How big the index ? Are there any other error messages in the TrexIndexServer or TrexQueueServer trace?

Best regards,
Mikhail

[Former Member]

Hi Mikhail,

This is the continuous messages on the TrexIndexServer or TrexQueueServer trace. And the trace files are huge in size, result of which we have to weekly remove the files to gain space.

The Trex version that we are using is 7.10.41.00.

Regards.

[former_member217429]

Hi Ambarish,

the TREX version is very old. Most likely the corruption of the index is the result of the bug in
TREX functionality. Could you please update the TREX to the REV 63 and recreate the index.Is it Windows or Linux installation? Could you please also check if you have any 3rd party tools that
have access on the TREX data - note 808768.

Best regards,
Mikhail

[Former Member]

HI Mikhail,

It is linux installation.

We are not using a third party tool.

Regards,

Ambarish

[former_member217429]

Hi Ambarish,

what's about the size of this index? Could you please check the size of the index as well as dictionary.read and dictionary.write files.

Best regards,
Mikhail

[Former Member]

HI Mikhail,

I am not sure as to how to check the index size?

Can you guide?

Regards.

[former_member217429]

Hi,

you can check the index size using the TREXAdminTool  (or tc. TREXADMIN) on the Index Admin tab. As fot the single files you need to check the size of dictionary.* files on the file system level directly using ls -l or ll command.
What's the Linux release you use? Is it probably SLES 11 SP 2 or 3 ?

Best regards,
Mikhail

[Former Member]

Hi Mikhail,

The screenshot is attached, the traces which are giving errors are blank in TrexAdmin. Rest are like 1.76 KB and so on.

The verison of linux is Red Hat Enterprise Linux Server release 6.4.

Regards.

[Former Member]

This might be the reason the errors are prompting as show cannot determine index server for index xi_msg:af.sid.hostname.

This error is already attached.

[former_member217429]

Hi,

no, this is not the root cause of the issue , but one of symptoms that the index is corrupted.

I would like to ask you to update the TREX to the newest REV 63 and recreate the index.
Best regards,
Mikhail

[Former Member]

Hi Mikhail,

Let me check with it will update you once I complete the activity.

Regards.

[Former Member]

Hi Mikhail,

Found out that the indexes were in status red. Deactivated and reactivated the indexes.

Now in the trace getting the following error after activating, though the status shows green but still the error stays:

An error occurred:2) Class: com.sap.aii.rwb.web.trex.CentralTrexException : Message: TrexException in Method: CentralTrexManager: getDocumentCounter( String ). TrexException: Failed to return the number of indexed documents for Trex index: xi_msg:is.00.hostname. Message: Class: com.sap.aii.af.service.trex.TrexException : TrexException in Method: TrexManager: getDocumentCounter( String ). Failed to return the number of indexed documents for Trex index: xi_msg:is.00.hostname. Message: Class: com.sap.aii.af.service.trex.TrexException : The index is temporary not available;index=xi_msg:is.00.hostname ;

Regards.

[former_member217429]

Hi Ambarish,

is it possible for you to open the OSS incident for this issue. I suppose that we could have some other corrupted indexes that cause the restarts of the TREXIndexServer process.

Best regards,
Mikhail

[Former Member]

Hi Mikhail,

The indexes are all green now, but the error prevails.

The other interesting point is our PRD server is not pointing to our DEV Trex, but still we see that the alerts from PRD are appearing in the trace files.

We were having alerts from QAS server too, the issue was same, the indexes were in de-activate mode. Re-activated and now there are no alerts from QAS.

Only DEV because of the error and PRD which is not maintained.

I am not sure on the OSS part.

Regards.

[former_member217429]

Hi ,

I suppose that you mean the index status on the Index Landscape tab of TREXADMIN. In your
case the status just means if the index is loaded or not. However from the TREXAdmin it's not possible to identify if the index is corrupted or not . The data can be inconsistent so that no indexing
calls can be executed on this , however the index can be loaded and used for the search (the quality
of the search results is dependend on the structure of the query and the status of the single binaries in index).

Best regards,
Mikhail

[former_member223839]

Hi Ambarish,

From the following error :

Error:Document status was not returned from the server.

it looks like there is no proper handshake, could you check whether your RFC is working correctly?

Regards

Amit

[Former Member]

Hi Amit,

The RFCs are working fine.

Regards,