Re: Re[2]: Error 208: Shared memory problem

>Well, this sounded interesting so I did: onmode -I 208. It didn't appear >to do anything, but I just noticed this in my message log: > >10:20:51 TrapError set, errno = 208, session_id = -1 What this means is that an af.xxx file will be produced when a 208 error is about to be returned. This will give us the stack trace which can be used to determine the nature of the problem. > >I think it means that when I get a 208 error, an af.xxx file will be >produced. > >So, how do I turn it off? Bounce the db? From what I remember - try onmode -i 208, however I normally just bounce the engine. > >Thanks, Dianne > > >In article <19970913234200.TAA11796@ladder01.news.aol.com>, SaTriGuy ><satriguy@aol.com> writes >}>I am using a newly configured Informix install on a RISC running >}>AIX4.1.2 and every time I try to join mor than three tables in an SQL >}>statment, I get an error 208. >}> >}>I have spent close to 20 hours on the phone with five different tech >}>support folks at Informix and none of >}>them were able to help me or >}>reproduce the problem. I have the same code running on another machine >}>without the error. >} >} >}If this is a 7.2 installation, you can try the following - >}onmode -I 208. This will produce an af.xxx file containing the stack when >}the error occurs. >} > > Is this documented - I cannot see it in the 7.2 Online Admin Guide. No --- It is not documented. It is one of the support-type keyins that was introduced in 7.2 > >}Also, if you are able to reproduce the problem at will, a SQLIDEBUG >}session might help to determine if the error is being generated on the >}engine or client side. >} > How does this work? Basically it creates a binary file of the low-level communication between the engine and the client. It can be useful to isolate this type of problem, but can produce a rather large file. That's why we normally want a re-creatable test case before using it. > >}Finally, if the front-line engineer is unable to provide the support that >}you need, simply request to speak with their manager. >} >}You could be running into a configuration issue, a physical memory >}limitation issue, a bug, or somthing else. >>-- Madison Pruet