Errors when using Re-Index

Squiz Matrix Support
1

I just set up a search page, and then tried to re-index my entire site, or even just a root node. The Hipo seems to hang and not make any progress, trying to process that first item and then finally gives up, here are the errors:

    MySource Warning
    HIPO Server Error : No TaskID Supplied or invalid TaskID [HIPO0048]
    
    MySource Warning
    Forcing Abort of HIPO Job [HIPO0046]


And the HIPO Error Report:

    ERROR:  [ASSERT EXCEPTION] DB Error: constraint violation INSERT INTO sq_sch_idx ( value, type_code, type, assetid, component, score ) VALUES ( 'college', 'folder', 'text', '466', 'metadata:254', 1 ) [nativecode=ERROR: duplicate key violates unique constraint "sch_idx_pk"] (LINE 700 IN [SYSTEM_ROOT]/packages/search/search_manager/search_manager.inc) [SYS0270]
    Reported 14th Mar 2007 10:44am


Do I just have some configuration for the Search wrong? Strange that the Hipo is talking about a "duplicate key violates unique constraint" sounds strange.
2

What version of Matrix are you running? There shouldn't be any more duplicate key errors occuring with the latest versions.

3

3.12.0 SSV

4

[quote]3.12.0 SSV
[right][post=“15137”]<{POST_SNAPBACK}>[/post][/right][/quote]



Can you log a support request for this? I want one of our support guys to check your sq_sch_idx table’s primary key to see why its throwing a duplicate key error.

5

Will do. Thanks.

6

There were a few (two I think) problems with search indexing. I think one was special character related and the other was a problem with words like +280 and -280 being treated the same.


These have only been recently discovered and fixed, so will not be released until 3.12.3 next month. Maybe you are suffering from one of these problems.

7

Looking at that SQL line, its a metadata value on the folder that's throwing the duplicate. We'd have to see what the values for asset ID 466 are to see if that's the problem.

8

Hmmm....if this is the case I hope that Squiz support does not use my support hours if this is the case, maybe we should just wait for the update...

BTW, asset 466 is a Position Formats folder.
9

[quote]BTW, asset 466 is a Position Formats folder.
[right][post=“15147”]<{POST_SNAPBACK}>[/post][/right][/quote]



Interesting. Can you check the metadata assigned to this folder? That’s where the error is being generated.

10

It just has the general site metadata schema applied to it which gives meta for site description, keywords, copyright, etc. Not sure what else to be noticing.

11

[quote]It just has the general site metadata schema applied to it which gives meta for site description, keywords, copyright, etc.  Not sure what else to be noticing.
[right][post=“15151”]<{POST_SNAPBACK}>[/post][/right][/quote]



If you remove the metadata schema from that asset, does it allow you to index correctly?

12

Nope. I even removed it for the whole site. HIPO hangs on Reindexing Position Formats and gives the same error.

13

[quote]Nope.  I even removed it for the whole site.  HIPO hangs on Reindexing Position Formats and gives the same error.
[right][post=“15160”]<{POST_SNAPBACK}>[/post][/right][/quote]



If it hangs with the same error, then metadata isn’t being removed correctly – that error is coming from the search index attempting to index the metadata for that asset. Can you verify that the metadata screen for that asset now has no content?

14

[quote]If it hangs with the same error, then metadata isn’t being removed correctly – that error is coming from the search index attempting to index the metadata for that asset. Can you verify that the metadata screen for that asset now has no content?
[right][post=“15162”]<{POST_SNAPBACK}>[/post][/right][/quote]



Spotted this question while browsing the forum - just to add that I’ve also logged a support ticket for a remarkably similar problem - I believe our reindex is also hanging at a Position Formats folder, and Metadata regeneration also hangs. I may try this workaround.



Could you please post a link to the relevant bug reports? We are using 3.10.5.

15

I had to remove the design area from my parse file, and also take off another metadata schema applied to my news items. After I did this it FINALLY indexed.

Why would my metadata be making the index not work? Sounds like the bug.
16

[quote]I had to remove the design area from my parse file, and also take off another metadata schema applied to my news items.  After I did this it FINALLY indexed.
[right][post=“15173”]<{POST_SNAPBACK}>[/post][/right][/quote]



I don’t think the design area would make a difference, but removing the second metadata schema may have been the issue (if that schema was also applied to the folder in question). How big is a backup of your entire system? It might be good for us to get a copy so that we can check this out.

17

I put in a support request for this, so Squiz Support is looking into the issue. Right now the site is super small, not much content has been added as this is a new site (I had set up the search after a few assets were created, so I needed to re-index). I have just been starting to build in a lot of the dynamic functions we will need, and have not added much real content. So a backup would be super small.
18

[quote]So a backup would be super small.
[right][post=“15179”]<{POST_SNAPBACK}>[/post][/right][/quote]



Could you put this backup tarball in a web accessible location and PM me the URL? I’ll grab a copy so that I can poke and prod it when I have some time.