cancel
Showing results for 
Search instead for 
Did you mean: 

EV 10 Index ranges displaying odd behaviour, ever increasing Journal items waiting index

EVSpinner
Level 5

Scenario

 

EV 10.0.4

Discovery Accelerator failing to return results for a period occupying last 12 months

The focus of the search is aimed at the Journal

Checking the Journal indexes, in Index mananer I see many indexes with ranges ending - Not Set, so I go to Sync. No impact

I chose one Index and kicked off a rebuild, and get the following after about 10 seconds:

 

Rebuild index volume started at 04/11/2013 17:14:18

 

Archive Name: PPS-XCMB-CLS Journal

Archive Id: 102188681D4D6F0459065759EB62E91841110000vault.bfl.local

Rebuild Scope: Index Volume

Index Volume Identity: 10122

Index Volume Range: 7891744-

Root Path Id: 1116092EAB3D5EF418F898566CB4875131810000vault.bfl.local

Folder: 102188681D4D6F0459065759EB62E9184_10122

Root Path: I:\Indexes\index7

Index Server: EVIndex01

Volume Type: 64-bit

 

04/11/2013 17:14:19 The index is empty, re-creating the collection to ensure that the collection is valid.

04/11/2013 17:14:19 Deleting the existing collection.

04/11/2013 17:14:21 Recreating the collection.

 

04/11/2013 17:14:22 The rebuild has completed successfully and reported no issues.

 

All the time the VAC reports Journal Items Awaiting Index increases - 3m and counting

I've DTraced Index Broker and Server - both indicate Indexing is working (or perhaps better to say, it's not failing)

I've checked the Journal DB fragmentation, and I've got references to Journal tables - 80%-90% Extent Scan Fragmentation - not good

I've got a DBA to look at the SQL Maintenance plan and attend to this fragmentation, but I guess my question is - am I on the right track in terms of troubleshooting

If the tables are fragmented so much that it is causing performance problems, how does this relate to the Index ranges 'No Set'. How come they don't do what they normally do when I rebuild - reset to 0 and then startuing slowly building up again?

1 ACCEPTED SOLUTION

Accepted Solutions

EV_Ajay
Level 6
Employee Accredited

Hi ,

Regardless for rebuilding the Index Volumn , i will recommende to to the EV Browser Search against those Index Volumn who range "Not set" and then check the Index Index Range.

 

If SQL fragmentation goes beyond the 80% then SQL will not able to create new index for table and unable to write data. Hence SQL fragmentation is require.

 

View solution in original post

3 REPLIES 3

EV_Ajay
Level 6
Employee Accredited

Hi ,

Regardless for rebuilding the Index Volumn , i will recommende to to the EV Browser Search against those Index Volumn who range "Not set" and then check the Index Index Range.

 

If SQL fragmentation goes beyond the 80% then SQL will not able to create new index for table and unable to write data. Hence SQL fragmentation is require.

 

EVSpinner
Level 5

Hi Ajay, I wrestled with a DBA today who said "My opinion is that Extent fragmentation is not and never will be an issue", but they ran SQL maintenance plan anyway and reduced the fragmentation to 15%

Now the items waiting are dropping at a rate of 60k an hour

EV_Ajay
Level 6
Employee Accredited

Hi,

Thanks for marking the solution. After SQL maintence the item awaiting count is reduce.