Atomikos Forum

Hello,

I think I have a similar case as the one described in default82df.html?community.6.1530
and
default82df.html?community.6.1524.0.

However, for me this also happens for XA Datasources (unlike the problem described in the 2nd link).


I found this problem when I simulated an error in the commit phase. I did this by adding a "throw new XAException()" in
XAResourceTransaction.commit() before the code " xaresource_.commit ( xid_, onePhase );":


As expected, Atomikos re-tries to commit this data to the database. However, when I stop my appserver and restart it, Atomikos does NOT retry to commit the data, but it just keeps logging the following error:

Exception on timeout of coordinator CNT0000300051: null

I think this a very similar error as the one described in default82df.html?community.6.1530. Do you already know when this bugfix will be released?

However, for me, the solution described in that link did not work, because the error does not occur in CoordinatorStateHandler, but in its subclass HeurHazardStateHandler.
The line

boolean committed = getCommitted ().booleanValue ();

in the onTimeout() method of HeurHazardStateHandler throws a NullPointerException because getCommitted returns null.


best regards
Stefan Mader

Hi,

Looks like the same thing indeed. This is currently our number one backlog item, so it should be in the next bug fix release.

Thanks

Stefan,

We are currently working on this issue. Quoting your comment above:

"However, for me, the solution described in that link did not work, because the error does not occur in CoordinatorStateHandler, but in its subclass HeurHazardStateHandler."

-> did you actually try? According to my analysis the suggested solution should work in your case too?

Thanks

Hi Guy,

I only made the corrections suggested in this other post, but I did not make any changes in HeurHazardStateHandler.
So I suppose that if you make similar corrections in HeurHazardStateHandler, this will probably solve the problem.

best regards,
Stefan