Conference iosg::all-in-1_v30

1438.0. "system-f-devoffline error in oa$mti_err" by KERNEL::OTHENJ () Wed Sep 16 1992 20:19

Hi,

	This is a highly political problem, so if anyone could come up with 
some definitive answers, I would really appreciate it.

vms v5.4 allin1 v2.4 patched up to K603 (wps v3.1)
Concurrent sender and fetcher
No Diskquotering                                                                                                                   
TRM not being run, but RSF is on a regular basis.                                                                   

	The customer had two very large (>800,000) SDAFs, and had a problem 
where one mail message was sent to three local users instead of one. This 
is really a seperate issue (and is being dealt with), but since then the 
customer has opened two new SDAFs, closed the others, and wants an explanation 
of any errors that arrive in oa$mti_err. Yesterday, he received the error

system-f-devoffline, device is not in configuration or not available                                                                

in oa$mti_err (the only error there). He has seen this error before, but it 
did not occur at the same time as the lost mail from the previous problem. 
The customer needs to know an explanation as to why this error would occur, 
apart from the obvious (the device was not off-line at the time).

	I have read the previous notes (827 and from the previous notes 
conf), but there does not seem to be any conclusion as to which situation 
caused the problem. Here are the answers I recieved from the customer from 
the previous notes entries;

1) PC send mail and power down before cursor back - Yes, this may have 
happened.
                                                                             
2) Second-class to Paper mail, but invalid printer in profile.dat (check for                                                        
sender and postmaster) Yes - ALL their mail is sent second-class, so this 
could have been a problem.
                                                                                                          
3) File access timeouts on sdaf or pending - check I/O activity to the devices                                                      
using SPM. If I/O rate is over 1, the possibility of file timeouts occur. More                                                      
chance of this happening if RSF not run. May also be worth looking at IO rates                                                      
of other devices using same HSC, if other devices have higher priority                                                              
(hardware), then ALL-IN-1 disk may have to wait. Customer will check as                                                             
monitoring already in place.                                                                                                        


The customer has checked his shared dir's, and they all point to the 
correct location. 
SDAFs on different disks, volume shadowed. Machine is cluster consisting                                                      
of 6540,8650, and two 8800.

	Has anyone got any further to the reason for this error?

		Thanks for any help,

		Julie
    
                

Is there any possibility this is happening when a user is sending EXPRESS mail
but the terminal port becomes DISCONNECTed somehow?  It might be the user's
terminal that is "offline", not a disk...

/Marty

    Hi,
    
    	Thanks for the reply, but ALL mail is sent Second-class (company
    decision), so I doubt if this would be the case.
    
    
    	Julie

    Hi there,
    
    Could it be that the run ALL-IN-1 in a terminal emulator
    from there PC's and switch them off without terminating
    there emulator properly.
    
    Sender tries to broadcast new mail to this user, cannot find the 
    terminal device and signals DEVOFFLINE.
    
    Just a thought.
    
    Regards,
    
    Jan
    

    
    	I think the reason you are not getting much response to this
    	one is that is was discussed in the old conference and we didn't
    	have much idea what it was then either. Remember, this is an
    	informal support channel....
    
    Regards,
    Paul

    I have a customer with this problem who says that SPM reports an i/o
    rate > 1 on the disk where the shared area concerned is. This may or
    may not be the problem?
    
    Anyway, does anyone have any reports of this in v3 ? (I imagine it's
    still a problem as the engineers aren't sure what causes it.)
    
    Thanks
    
    Clive