Running a report without indexes

[Kenneth Brody]

On 3/19/2013 1:55 PM, Ian Wood wrote:
> This is the command line
>
> \fp\dreport gr_ap -fp file_dump -u -a -v file_dump_s -y none -ro -r "3CON"
> -h "gr_ap"
>
> No index is specified, and I can assure you that there are no lookups at
> all in either the selection process (which does contain a call, but that's
> only to set a date, again no lookup) nor the output process.
>
> I guess I was just looking for a PFNOINDEX=ON or something of that ilk so
> as not to even try to read any indexes.

Given the situation you describe, filePro should only open the automatic 
indexes when it opens the "gr_ap" file, and not touch them again until the 
file is closed.

If you run the same report on the same data, do you get the same error at 
the same place?

There is no way to "turn off" indexes.  If you update a record, any 
automatic indexes built on fields whose value has changed *must* be updated. 
  And, if you use an index for a lookup, then obviously filePro *must* use 
that index.

> On Tue, Mar 19, 2013 at 12:57 PM, Kenneth Brody <kenbrody at spamcop.net>wrote:
>
>> On 3/19/2013 10:50 AM, Ian Wood wrote:
>>
>>> Last night's error was on my A/P file, index.M, there's no consistency on
>>> which file or index it occurs on.  There are no lookups being performed,
>>> it
>>> is a straight export ascii dump of the data, it is the exact same code in
>>> at least 35 of the files.
>>>
>>
>> Because you are running with "-ro" and not modifying anything, unless you
>> are running the report off of index M, I can think of no reason why filePro
>> would be accessing that index, aside from a lookup.
>>
>>
>>   When I went into my A/P file this morning, index.M does not say it's an
>>> invalid index, and accessing the file via that index appears to be all in
>>> order.
>>>
>>
>> You may not have accessed the corrupted part of the index.
>>
>>
>>   On Tue, Mar 19, 2013 at 10:42 AM, Kenneth Brody <kenbrody at spamcop.net
>>>> wrote:
>>>
>>>   On 3/19/2013 9:55 AM, Ian Wood wrote:
>>>>
>>>>   filepro 5.07.03.02
>>>>> Windows Server 2008
>>>>>
>>>>> We currently run an overnight process that exports data from about 50
>>>>> files
>>>>> to feed another system.  Periodically, we get an invalid index error on
>>>>> one
>>>>> of the files, not one in particular, just a random spot.  I have the -ro
>>>>> flag set to run as read only.
>>>>>
>>>>> Is there a way to have rreport run on files and ignore automatic
>>>>> indexes?
>>>>> I'm not posting anything, just doing file dumps.
>>>>>
>>>>>
>>>> If you're running with "-ro", then the problem isn't when updating an
>>>> index (which it won't do in read-only mode), but when reading an index.
>>>>    Assuming that the error doesn't occur on the temp index that *report
>>>> built, then I can only assume that the error occurred on a lookup.  (What
>>>> index did the error occur on?)  And the only index being accessed when
>>>> executing the lookup it the index being used for that lookup, which you
>>>> obviously can't avoid using.
>>>>
>>>>
>>>>    I don't have enough time to do a dxmaint -ra -e on all files first as
>>>> we
>>>>
>>>>> run two shifts and the exports take approximately 5-6 hours to run.
>>>>>
>>>>>
>>>> If you are looking up to a file via a corrupted index, you may have no
>>>> choice but to rebuild that index.

-- 
Kenneth Brody