[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Ftpapi] LIBHTTP/QRPGLESRC,COMMTCPR4



I recently started consuming several AWS api’s and I’m getting an error I’ve never seen before. I can’t seem to recreate the error and it always seems to happen in a different area of the process.  

 

I was able to narrow it down a little further by running in debug. It breaks in COMMTCPR4 at this procedure…

 

P CommTCP_FD_SET...                                                     

P                 B                   EXPORT                             

D CommTCP_FD_SET...                                                     

D                 PI                                                    

D   peFD                        10I 0                                   

D   peFDSet                           like(fdset)                       

D wkByteNo        S              5I 0                                   

D wkMask          S              1A                                     

D wkByte          S              1A                                      

C                   callp     CalcBitPos(peFD:wkByteNo:wkMask)          

c                   eval      wkByte = %subst(peFDSet:wkByteNo:1)       

c                   biton     wkMask        wkByte                      

c                   eval      %subst(peFDSet:wkByteNo:1) = wkByte       

P                 E                                                     

 

wkByteNo turns out to be 32.

It makes sense because peFDSet is only 28 bytes.

I’ve re-run the same request immediately after this error and it works just fine. If I happened to be running it interactively I have to sign-off then sign back on first.

 

We’ve considered increasing the length of the fdset variable in LIBHTTP/PRIVATE_H but there are so many procedures referencing it. I don’t want to rock the boat.

I also have a hard time believing that the variable in the copybook needs to be changed. It HAS to be something I’m doing right?

 

The only real difference I can spot is that this new AWS stuff requires a lot more http requests per process than any other web service we use. We’re talking 3-4 times more. Could the amount of requests in a job be related?

 

Thanks

Anthony Shrader

Maples Industries, Inc.

ashrader@xxxxxxxxxxxxxx

256-259-1277   Ext:649

Description: MaplesLogo2

 

-- 
_______________________________________________
Ftpapi mailing list
Ftpapi@xxxxxxxxxxxxxxxxxxxxxx
http://scottklement.com/mailman/listinfo/ftpapi