Cannot kill many processes whose ppid is 1

modifosi · 04-21-2020, 05:08 AM

Hello.
In a production system SunOS we found the same processes witth ppid = 1, so there must be going on with the application and there are more than 3k processes and increasing, for instance

Code:

ps -fu xpinvoice| grep launch_web|wc -l
    3086

Code:

bash-3.2$ ps -fea | grep launch_web
xpinvoice 56525     1   0 15:00:29 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 68873     1   0 12:08:37 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 15141     1   0 14:36:50 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice  6962     1   0 12:10:16 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 69284     1   0 16:31:45 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 37469     1   0 15:25:49 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 48415     1   0 12:33:02 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 47245     1   0 12:18:31 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 62592     1   0 12:07:21 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 66140     1   0 16:46:16 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 15841     1   0 15:50:22 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 11251     1   0 14:21:22 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG
xpinvoice 26343     1   0 15:08:44 ?           0:00 sh //produccion/explotacion/xpinvoice/facturacion/ksh/launch_web.ksh -u/ -l3 -eFG

I could not kill those processes, so i escalete up this issue to our sys administrator who says there is no way to kill these processes because they are tied to the kernel, so it is necessary to restart the server, so these question come up:

1 its kind of odd to find the same processes with too many processes with ppid =1 ? so it has something to do with an application failure?

why did our sys admin say these processes are related to the kernel and its necessary to restart the server?

I appreciate your help in advanced?

berndbausch · 04-21-2020, 07:07 AM

Quote:

Originally Posted by modifosi

its kind of odd to find the same processes with too many processes with ppid =1 ? so it has something to do with an application failure?

That would be my first thought. Most probably, the parent(s) of these processes died, and they were adopted by init.

Quote:

why did our sys admin say these processes are related to the kernel and its necessary to restart the server?

These are shell processes. I don't know Solaris or SunOS, but I can't imagine that shell processes run in the kernel. Your sysadmin is almost certainly wrong.

What happens when you try to kill them? Such as kill -9 6962?

Is there nobody in your datacenter who knows what application that is and how to troubleshoot it? Are there any log files?

EDIT: The // at the start of the filenames could indicate that the scripts are located at a remote server named produccion, which is perhaps not accessible. The processs might be waiting for high priority IO, which could make them unkillable. A lot of speculation, ifs and perhaps, though.

EDIT2: Wikipedia tells me that the most recent release of SunOS was in 1994. Are you certain this is SunOS?

ehartman · 04-21-2020, 09:52 AM

Quote:

Originally Posted by berndbausch

Wikipedia tells me that the most recent release of SunOS was in 1994. Are you certain this is SunOS?

It's called Solaris (and openSolaris) nowadays.

rknichols · 04-21-2020, 12:59 PM

What is the "ps -l" output for a few of those processes? The only "unkillable" condition (where even "kill -9" is ineffective) I'm aware of is a process stuck in the "D" state, waiting for some I/O operation that never completes.