Bug 63463

Summary:	invasion of kio-slaves. All called kio-slave remain present and acive, although their calling app/process already has been terminated.
Product:	[Applications] konqueror	Reporter:	Detlef <djbloch>
Component:	general	Assignee:	Konqueror Developers <konq-bugs>
Status:	RESOLVED FIXED
Severity:	normal
Priority:	NOR
Version First Reported In:	unspecified
Target Milestone:	---
Platform:	openSUSE
OS:	Linux
Latest Commit:		Version Fixed In:
Sentry Crash Report:

Description Detlef 2003-08-30 15:50:28 UTC

Version:            (using KDE KDE 3.1.2)
Installed from:    SuSE RPMs
OS:          Linux

Keeping the KDE system active for about a day, without rebooting it, and unsing its mail and browser facilities, the system get slower and slower in its responses; e.g. at the machine, not from remote, using an editor, it takes time to see typed in characters being echoed on screen. Using "ps x" shows a very large number of kio-slaves being present. Thus removing them "manually" using e.g. ksysguard I'll have this situation:

  PID TTY      STAT   TIME COMMAND
...
 2038 ?        S      0:01 kdeinit: kio_uiserver    
 2224 ?        S      1:54 ksysguard
 2225 ?        S      1:33 ksysguardd
 2410 ?        S      0:01 kdeinit: konsole         
 2411 pts/6    S      0:00 /bin/bash
 2418 pts/6    R      0:00 ps x

Now opening a konqueror with 2 tabs quering a man-page through the appropriate kio-slave creates the following:

  PID TTY      STAT   TIME COMMAND
...
 2410 ?        S      0:02 kdeinit: konsole         
 2411 pts/6    S      0:00 /bin/bash
 2420 ?        S      0:11 kdeinit: konqueror --silent
 2421 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorvSuiib.slave-socket
 2441 ?        S      0:00 kdeinit: kio_about about /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorJ0uzWa.slave-socket
 2442 ?        S      0:00 kdeinit: kio_man man /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konqueror30FOja.slave-socket
 2444 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorwo23cb.slave-socket
 2445 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket 
/tmp/ksocket-kreti/konquerorGhrHya.slave-socket
 2447 pts/6    R      0:00 ps x

Closing the Konqereror creates this:

  PID TTY      STAT   TIME COMMAND
...
 2410 ?        S      0:02 kdeinit: konsole         
 2411 pts/6    S      0:00 /bin/bash
 2421 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorvSuiib.slave-socket
 2441 ?        S      0:00 kdeinit: kio_about about /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorJ0uzWa.slave-socket
 2442 ?        S      0:00 kdeinit: kio_man man /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konqueror30FOja.slave-socket
 2444 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorwo23cb.slave-socket
 2445 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorGhrHya.slave-socket
 2450 pts/6    R      0:00 ps x

Repeating the above procedure: Open Konqueror, creating some tabs, man-paging and closing Konqueror results in:

  PID TTY      STAT   TIME COMMAND
... 

 2410 ?        S      0:03 kdeinit: konsole         
 2411 pts/6    S      0:00 /bin/bash
 2421 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorvSuiib.slave-socket
 2441 ?        S      0:00 kdeinit: kio_about about /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorJ0uzWa.slave-socket
 2442 ?        S      0:00 kdeinit: kio_man man /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konqueror30FOja.slave-socket
 2444 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorwo23cb.slave-socket
 2445 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorGhrHya.slave-socket
 2451 ?        S      0:39 adie
 2474 ?        S      0:00 kdeinit: kio_about about /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorToo63b.slave-socket
 2475 ?        S      0:00 kdeinit: kio_man man /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerorWCJflb.slave-socket
 2476 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konquerororpV2a.slave-socket
 2477 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konqueror0GbDOa.slave-socket
 2478 ?        S      0:00 kdeinit: kio_file file /tmp/ksocket-kreti/klauncherUL2bSb.slave-socket /tmp/ksocket-kreti/konqueroryigvwa.slave-socket
 2480 pts/6    R      0:00 ps x

As you will see, always new slaves are created for their specific task, but fail to be removed, once the job is done. In http(s) situations, the memory gets flooded with their respective slaves, who all net manual attention to get to rest, or the system's responsiveness converges to zero.

Perhaps this is a configuration problem, but I can't find the proper tuning knob nor switch.

Thanks for your attention

Comment 1 former user 2004-04-05 14:31:54 UTC

IMHO kio_slaves not behaving corectly is the major bug also having much to do with ongoing  bug reports complaining about not being able to unmount removable filesystems 
(i.e. #37780 long! #76550 #53848 etc.)

Seems better to fix this general problem than hacking mount point specific workarounds (#38778).

Comment 2 Detlef 2004-05-13 11:10:03 UTC

20040513 Since KDE 3.2 with SuSE 9.0 as well as SuSE 9.1 this problem no longer occurred. For this reason I propose to close this Bug #.

Detlef

Comment 3 Stephan Kulow 2004-05-13 17:33:28 UTC

*** Bug has been marked as fixed ***.