[daip] AIPS startup and run problems

[Eric Greisen]

Sayali Avachat wrote:
> Hi, 
> 
> I have sent you another email including the the problems with AIPS after 
> we upgraded our system to Fedora 15.
> But I wanted to ask about one other problem which I encountered a couple 
> weeks before the system upgrade. I am asking this in different email 
> just to avoid confusion.
> 
> I was using CALIB for the self calibration of my data. After using P and 
> A&P self calibration runs I tried to use GCON with the proper GAINERR (15%).
> When I set SOLMODE to GCON, it ran the task, almost finished but before 
> it could produce the final CALIB file, it just stopped and did not do 
> anything further. I don't remember the exact step at which it stopped, 
> but its the step just before it gives the final number of solutions 
> found and number of solutions it failed on, the created file number etc.
> 
> This task was running overnight once and did not let me use any other 
> further task. It was running in background all the time, until I aborted 
> the task.
> This was something that never happened to me before. 
> 
> Do you have any idea about such type of error? Any help will be appreciated.
> 
> If you want to know anything more in order to troubleshoot this, do ask 
> me. I will give you information how much ever I can remember.
> As I said in my other email, we are trying to re-install AIPS, currently 
> we are working on 31Dec2010 version. I hope it all works well after 
> re-installation

I really have little idea bout this question:

1. GCON mode is rarely used - I had to go see what it is about.  Most 
interferometers have similar telescopes and so wide variations in noise 
levels and gain variations between them are not expected and GCON would 
be mostly about that I suspect.

2. I do not see how CALIB could keep you from using any other task.  It 
is possible to start (GO) with DOWAIT TRUE for any task and then AIPS 
does not resume until the task quits.  But you can still run other aips 
numbers in other xterms.  It is possible for CALIB when computing models 
  to use a large fraction of the cpu but that normally does not lock up 
machines.

I suppose that this method may have bugs that have not been found since 
it is so rarely used.  I will see if I can check on this...

ERic Greisen