[RAS] [External] Re: RAS database corruption

Christian Zimmermann zimmermann at stlouisfed.org
Sat Sep 22 00:42:07 UTC 2018


Repeatedly recreating the ACIS database, I finally managed to get to be 
OK. The records are still invisible to users.

I repeated this for the RePEc database, same result.

This is really strange. I launch updareq on the problematic series, the 
logs show that the records are considered to be there, thus not added. So 
maybe there are indeed there, but for some reason not served.

Christian Zimmermann                          FIGUGEGL!
Economic Research
Federal Reserve Bank of St. Louis
P.O. Box 442
St. Louis MO 63166-0442 USA
https://ideas.repec.org/zimm/   @CZimm_economist

On Wed, 19 Sep 2018, Christian Zimmermann wrote:

> The ACIS databases have been completely refreshed, the files db now passes 
> the tests, the records db now has a large number of "overflow item 
> incomplete" errors and fails.
>
> RAS is still functional, but the records that were invisible to users are 
> still so, so nothing has been resolved.
>
> Christian Zimmermann                          FIGUGEGL!
> Economic Research
> Federal Reserve Bank of St. Louis
> P.O. Box 442
> St. Louis MO 63166-0442 USA
> https://ideas.repec.org/zimm/   @CZimm_economist
>
> On Thu, 13 Sep 2018, Thomas Krichel wrote:
>
>>   Christian Zimmermann writes
>>
>>>  RAS has a corrupted database. RAS still runs, but some series have become
>>>  invisible to it and authors complain their profiles got emptied from
>>>  those
>>>  items. Trying to push them back (updareq) gets a response that they are
>>>  already there, so nothing is done about it.
>>>
>>>  I tried to recover. It exits with no troublesome message, but verifying
>>>  the
>>>  database gives issues in ACIS/files and ACIS/records. I tried fixing them
>>>  from dump, but dumping crashes. Backups have even more errors when
>>>  verifying
>>>  them.
>>
>>   I suggest you make sure you use recent versions of mariadb berkeleydb.
>> 
>>>
>>>  At this point, I think I am left with having to rebuild the database. But
>>>  I
>>>  find no indication on how to do that.
>>
>>   I think you need to empty the database of records, then move the Berkley
>>   files out of the way and then start updareq. I would take RAS offline
>>   while this is done.
>>
>>  -- 
>>
>>   Cheers,
>>
>>   Thomas Krichel                  http://openlib.org/home/krichel
>>                                               skype:thomaskrichel
>> 
>



More information about the RAS-run mailing list