[Glass] Mixing Unicode and String instances in Indexes [WAS] Re: Further commits have been disabled for this session because: 'CorruptObj error'. This session must logout.

[Dale Henrichs via Glass]

On 10/28/2015 06:47 PM, Mariano Martinez Peck wrote:
> OK Dale, I think I found out what is going on.
>
> Please, read carefully steps 1 to 3 from my original email. With that 
> in mind, the problem is this. I run this code:
>
>
>
> | collectionsWithBadIndexes |
> "Run a query so that to be sure we use indexes"
> Object new assert: ((UserGlobals at: #IndexesBugReport) select: { 
> :each | each.testSelector = 'aString' }) size = 1.
> "Check health of indexes"
> [[
> IndexManager current removeAllIncompleteIndexes.
> collectionsWithBadIndexes := IndexManager current nscsWithBadIndexes.
> Transcript show: 'There are ' , collectionsWithBadIndexes size 
> asString, ' collections with bad indexes.'; cr.
> Transcript show: 'The following are the OOPs of such collections: '.
> collectionsWithBadIndexes do: [ :each | Transcript show: ' - ' , each 
> asOop asString.  ].
> Transcript cr.
> ] on: Error do: [ :err | Transcript show: ' Original error: ' , err 
> description; cr. ]
> ] ensure: [ System commit ]
>
> It seems that #nscsWithBadIndexes let things in a very bad state. The 
> original error is the
>
> /" Original error: a InternalError occurred (error 2261), The object 
> with object ID 20 is corrupt. Reason: 'relops: encrypt (no Unicode 
> collator) -- Unicode strings encountered'/
> /
> /
> /AUTOCOMMIT DISABLED"/
>
> And the one trying to do the last commit is the
>
> /"'a TransactionError occurred (error 2249), Further commits have been 
> disabled for this session because: ''CorruptObj error''."/
/Okay I think I discussed these things in my earlier response .../
>
> So..why did I get in seaside request?
Yes! ... this is still a mystery...
> I think because #seasideProcessRequestWithRetry:resultBlock:    tries 
> again and so, the problem.
> Why do I got  with loading Seaside from metacello? again, because it 
> tries 3 times (when it fails, it tries again).
> So it seems that the error "'a TransactionError occurred (error 2249), 
> Further commits have been disabled for this session because: 
> ''CorruptObj error''. This session m..."    comes when you try to 
> commit again after getting the error "'a InternalError occurred (error 
> 2261), The object with object ID 20 is corrupt. Reason: ''relops: 
> encrypt (no Unicode collator) -- Unicode ..."
> Maybe this is related to the "Autocommit disabled" ?
Yes Seaside will try to commit when encountering an error (to record 
things in the object log), but commit prohibiting errors don't allow the 
commit and you get to where the commit fails ...

And yes, Seaside continues to retry on commit failures and if somehow 
the ensuing requests don't trigger the index error you will be allowed 
to commit again
>
> As for Seaside I imagine I only got the last error and not the 
> original, and that's why I couldn't see the original stack.
>
> Does any of this make sense?
>

Yes ... and now I need to look into things for indexes to see if I can 
avoid issuing a corrupt obj error when you are doing read only 
operations ... I think my original logic, is that obviously the index is 
corrupt if the collator is missing, but I anticipate that in an upgrade 
scenario it will be "normal" to have a collator-less index containing 
Unicode strings ...

So will submit a bug ... I'm not sure that I will be able to safely fix 
this bug, but the bug report (and bug notes with wrokaround) are worth 
having ...

Thanks for you effort,

Dale
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gemtalksystems.com/mailman/private/glass/attachments/20151104/586aad72/attachment.html>