Hi, Twice now since installing 9.3 I've found an instance of fam (/usr/sbin/fam) soaking up all the CPU cycles it could get. This morning when I awoke was the most recent occurrence. When I noticed it, it had just passed the 8-hour mark of cumulative CPU time. I had remained logged in and do not lock the screen, though I have a screen saver configured as well as DPMS enabled. There was no trouble accessing the existing session. Other, probably related symptoms include hang of KMail during start-up (its window displays but does not get drawn and subsequently must be forcibly terminated) and Mozilla not getting far enough to even display a window before hanging. The fam process appeared to be owned by my UID, but neither I nor root could kill it ("Operation not permitted"). Attempting to renice it gave the same error message. I was able reboot without further incident (I was worried that the unkillable process would wedge the shutdown process) but afterward the "kdm is changed to xdm" problem reoccurred. Has anybody seen this or anything like it? Have you solved it, perchance? I have fam-2.6.10-123 installed. Thanks. Randall Schulz
Hello again, A follow-up: In hunting around on the Web for information on this problem with fam hanging, I found a connection with the use of k3b (http://lists.debian.org/debian-kde/2004/01/msg00253.html). In fact, last night before going to bed, I used k3b to write a backup CD. So it seems moderately likely that this is the trigger in my case, too. What's most odd is that this problem appears to go back over a year at least. This thread, http://lists.debian.org/debian-kde/2003/12/msg00144.html in the debian list seems to confirm the problem. Apparently Konqueror (in its local file system browser persona) appears to use fam, too, so perhaps the fact that it was running with four file system viewing tabs open was part of the problem, though I did not try to use it when I encountered this problem. Both of these discussions state that fam is optional. If this keeps happening, perhaps I'll try to get by without it. The other thing I can think of is that the odd state in which the fam process existed when I found it, being unkillable and not reniceable, makes me think that maybe the problem is in dnotify (a kernel-level facility used by fam, if I understand correctly). Another thing I just realized is that all those CPU cycles that fam was soaking up were kernel-mode (I configured GKrellm to make the difference more apparent), so perhaps the problem does lie outside fam itself? Unfortunately, FAM appears to be an SGI open-source project and I cannot find a bug-reporting facility there. Randall Schulz On Sunday 01 May 2005 07:26, Randall R Schulz wrote:
Hi,
Twice now since installing 9.3 I've found an instance of fam (/usr/sbin/fam) soaking up all the CPU cycles it could get.
...
I have fam-2.6.10-123 installed.
Le Dimanche 1 Mai 2005 16:26, Randall R Schulz a écrit :
Hi,
Twice now since installing 9.3 I've found an instance of fam (/usr/sbin/fam) soaking up all the CPU cycles it could get.
just for info, Mandriva (ex Mandrake) has switched to gamin (http://www.gnome.org/~veillard/gamin/index.html) in place of fam
participants (2)
-
Frederic
-
Randall R Schulz