[comp.unix.aix] question about ps

pete@bally.Bally.COM (Pete Gregory) (03/22/90)

Hi -

Sometimes, when I do a ps -el (that's /bin/ps), I get the following
two errors (and no ps output):

	swapread: dbd not of type DBD_SWAP
	getupages: Can't read upage from swap for proc nnnn

Any ideas?  The process number nnnn is a normal user process that is
there all the time, and not something that is just starting or exiting.

Thanks for your help......

Pete Gregory   : uucp:   uunet!bally!pete            | 
Software Engr. : domain: pete@bally.bally.com     ---|---
Bally Systems  : phone:  702-323-6156 x882           |         John 3:16
255 Bell St.   : FAX:    702-323-5997                |     (think about it!)
Reno, NV 89503                                       |

francis@hslrswi.UUCP (Francis Demierre) (03/24/90)

In article <237@bally.Bally.COM> pete@bally.Bally.COM (Pete Gregory) writes:
>Hi -
>
>Sometimes, when I do a ps -el (that's /bin/ps), I get the following
>two errors (and no ps output):
>
>	swapread: dbd not of type DBD_SWAP
>	getupages: Can't read upage from swap for proc nnnn
>
>Any ideas?  The process number nnnn is a normal user process that is
>there all the time, and not something that is just starting or exiting.
>
>Thanks for your help......
>
>Pete Gregory   : uucp:   uunet!bally!pete            | 
>Software Engr. : domain: pete@bally.bally.com     ---|---
>Bally Systems  : phone:  702-323-6156 x882           |         John 3:16
>255 Bell St.   : FAX:    702-323-5997                |     (think about it!)
>Reno, NV 89503                                       |


I have the same problem with an application I wrote that forks some 
process to test the printer (The forked process write to the printer and
wait the write to come back to know if the printer works or not. More
complicated than that but not relevant to the problem).

The thing is that while 'write()'ing, the same message comes up. The process
ID is random, sometime a normal user process, often the process ID of 
pciconsrv.ip (DOS Access server), etc....

The messages are described neither in the "Messages" manual nor anywhere 
else.

By the way, I'm running AIX 1.1 with update MMPP1402 installed.

If anybody knows about, I would be pleased to know about it too !

Thanks in advance, Francis.


-------------------------------------------------------------------------------
Francis Demierre       UUCP: ...!mcsun!chx400!hslrswi!francis
                       Unix mail: francis@hslrswi.UUCP
-------------------------------------------------------------------------------
-- 
-------------------------------------------------------------------------------
Francis Demierre       UUCP: ...!mcsun!chx400!hslrswi!francis
                       Unix mail: francis@hslrswi.UUCP
-------- Ha Ha Ha Ha Amiga !  Only Amiga can make computing so funny ! --------

jackv@turnkey.TCC.COM (Jack F. Vogel) (03/25/90)

In article <237@bally.Bally.COM> pete@bally.Bally.COM (Pete Gregory) writes:
>
>Sometimes, when I do a ps -el (that's /bin/ps), I get the following
>two errors (and no ps output):
>
>	swapread: dbd not of type DBD_SWAP
>	getupages: Can't read upage from swap for proc nnnn
>
>Any ideas?
 
What is happening here is nothing peculiar to AIX, it can happen on other
Unix variants as well as far as I know. It is likely to occur most often
under heavy system activity. I believe the scenario goes something like
this.... You fire up the ps command, it reads the process table and finds
that the user structure that it wants to read is swapped out. So it calls
swapread to get at it. Perhaps it sleeps on this I/O or is preempted for
whatever reason. Then during the time that it is not running, that process
that was swapped out it brought into core. Now when ps is again scheduled
the swapread fails because that page is no longer on the paging device.
As I said, I believe you should only see this with heavy system activity.
Also subsequent invocations of ps, due to the dynamic nature of things,
should read things successfully.

Hope this sheds some light.

Disclaimer: These opinions are mine, not necessarily LCC's

-- 
Jack F. Vogel			jackv@seas.ucla.edu
AIX Technical Support	              - or -
Locus Computing Corp.		jackv@ifs.umich.edu