I think you may be getting affected by this defect: SW00471616 Email
engine log shows exception "java.lang.NegativeArraySizeException"
*Hotfix Version Stamp:* 8.1.00 201410281446 Hotfix
*Files Patched:* emaildaemon.jar
armapi81_build001.dll
Check to see if you are getting this
"java.lang.NegativeArraySizeException" message
in your email engine log. I have seen this issue manifest, where once this
negative array size exception appears, email engine can start writing to
the wrong form.
You should be able to request the hotfix from BMC Support.
Hope this helps
On Tue, Jun 2, 2015 at 2:58 PM, Mckinnish, Randy <
randy.mckinn...@compass-usa.com> wrote:
> **
>
> Greetings List,
>
> I would like to pose this scenario to you all and get your take on it. It
> happened recently in our production environment. I want to share what we
> did to “fix” it but I would also appreciate the sanity check in case we
> totally missed something and could potentially be blindsided again later.
> We are running 8.1.02 on a load balanced pair of Windows (Server 2012 R2)
> boxes looking at a SQL (SQL 2012) database. Our email engine (MAPI)
> recently stopped working. This happened late on a Friday and so was
> unnoticed through the weekend until early Monday morning. When we began
> troubleshooting we saw some unusual references to field ID 450 and 451 in
> the stderr.log;
>
> ERROR (307): Required field can not be blank.; 451
>
> ERROR (307): Required field can not be blank.; 450
>
> Those field ID’s are unique to archive forms and we do archive email
> messages and have been since 6.x as we have some custom apps used by Legal
> and HR. We continued trying to rule out one server over the other by
> changing ranking but to no avail. We still had the error. We started doing
> compares between PRD and QAS and discovered that our ARCHIVE:AR System
> Email Message (our custom) in PRD no longer had its alias but now had the
> same alias as the AR System Email Messages form. We had recently run BPCU
> as the last step in our post 8.x upgrade and are theorizing that the alias
> change may have been a result of this. What we are struggling with is how
> it managed to be unaffected for a week before chaos ensued. The quick
> solution to get us back to normal was to migrate the email related forms
> with proper intact alias from QAS to PRD. The error started slowing and
> with a couple more restarts of the mail engine(s) we were back to normal.
>
> In closing, this isn’t a witch hunt for BPCU. It’s more a search for more
> ideas around what might have happened. Unfortunately in the heat of the
> moment, we never attempted to log SQL or API calls so we only have the
> email logs to go off of. In my understanding of the email engine and it’s
> configurability within the EmailDaemon.properties file there is no means to
> point it to another form manually so I have always assumed it is deeply
> embedded in the code. Now I question if that reference is a form name or
> simply the alias of the form assumed to never be disturbed or questioned or
> in our case, mistakenly mirrored causing the email engine to go with the
> “first available” resulting in the submit errors with the reserved ID’s.
>
> I’d appreciate any feedback or thoughts on the matter especially from you
> folks that have been out there for a while through numerous iterations of
> the platform. We did log this as a support issue at the time but “my
> evidence doesn’t support my analysis” and I have been asked to reproduce in
> a non-production environment. Not so sure I am willing to go there yet
> which is why I am shamelessly soliciting feedback from the list. Any takers?
>
>
>
> Have a great evening!
>
> Randy
>
>
> This email is subject to certain disclaimers, which may be reviewed via
> the following link. http://compass-usa.com/Pages/Disclaimer.aspx.
> _ARSlist: "Where the Answers Are" and have been for 20 years_
_______________________________________________________________________________
UNSUBSCRIBE or access ARSlist Archives at www.arslist.org
"Where the Answers Are, and have been for 20 years"
