Failed Journal Creation

> hi all
> 
> if anyone recognises this one I would be grateful if they could post the
> bug# (and patch# ?), as I'm trying to persuade CA that we have an actual
> problem.
> 
> symptom -
> - during online CKPs, the ii_acp.log will record that Ingres has updated
> the CNF file to reflect the creation of a new JNL file. This can be
> verified with infodb.
> - at the end of the CKP, Ingres will then try and create the new JNL file,
> but has failed 3 times in the last 2 wks. This causes DMFACP to stop !!
> 
> I can assure you all that it's *not* because someone is deleting the latest
> JNL file !! (I have better things to do at 5am, such as sleep ;-)
> 
> we're running 6.4/05 axpvms03 on a DEC Alpha 8400 (big kahuna!)
> 
> pls email (and post) any replies to wongjm@m1.com.sing (subst. "sg" for
> "sing")
> 
> TIA !
> Jeff
> 



Hi Jeff,

    In your case is there any chance that the journal disk was full. Under
    those circumstances the Archiver will certainly fail. Seeing as how
    checkpoints and journalsd are often kept on the same disk this is a real
    possibility.

    There are a few problems in 6.4 (UNIX) with the creation of Journal Files.
    Namely, not setting the next journal file number correctly (shows as
    zeroes in infodb output) or skipping a journal file completely. Both of
    these are associated with online checkpoints. Neither has ever stopped the
    Archiver. Although they can really screw a recovery!! Bet you can guess
    how I found that out.

    Even Assuming that you can show CA a new problem I'd be very doubtful of
    any activity on it. All they will say is upgrade to OI 2. Which probably
    isnt an unreasonable request.

    Martin Bowes
Ingres Q & A
Back to William's Home Page
Email William