Re: process stuck in DE state

Andriy Gapon wrote:

>The hanging application runs in linux emulation.
>
>$ ps ax -O wchan | fgrep 970
>970 fdlhol  p4  DE     0:00,09 strmqcsv WAS_node_si_server_sb4_si
>
>$ fstat -p 970
>USER     CMD          PID   FD MOUNT      INUM MODE         SZ|DV R/W
>mqm      strmqcsv     970 root /             2 drwxr-xr-x     512  r
>mqm      strmqcsv     970   wd /usr     1555303 drwxrwxr-x     512  r
>mqm      strmqcsv     970 text /usr     1536473 -r-sr-s---    7630  r
>mqm      strmqcsv     970    0 /dev         10 crw-rw-rw-    null  r
>mqm      strmqcsv     970    1 /dev         10 crw-rw-rw-    null rw
>mqm      strmqcsv     970    2 /dev         10 crw-rw-rw-    null rw
>mqm      strmqcsv     970    3* pipe c7076408 <-> c7af0158      0 rw
>mqm      strmqcsv     970    4* pipe c7af0158 <-> c7076408      0 rw
>
>ipcs -a | fgrep 970
>m  65561  286838271 --rw-rw-rw-      mqm      mqm      mqm      mqm
> 1   8192    970    974 18:47:11 18:47:11 18:47:11
>
>This happens from time to time, I can not reproduce this at will, so it
>looks like a concurrency issue. As I understand this happens in kernel
>fdfree().
>I use 5.2.1-release, so maybe it was fixed later, but I can not install
>-current (or 5.3 beta) at the moment.
>Is there anything I can do to avoid this problem and help debug it?
>


Unfortunatly a LOT of work was done in this area for 5.3.
We did make it clear I think that people loading 5.2 should be prepared 
to upgrade to 5.3 as
soon as it became viable because we expectedt too many problems with 5.2 
for us to go
back and fix it..

So Unfortunatly I'm going to suggest the one thing you said you really
 didn't want to have to do..
(WHY can't you go to 5.3?)



>
>  
>