| To: | nf2@xxxxxxxxxxxxx |
|---|---|
| Subject: | Re: fam using lots of CPU |
| From: | Jean-Marc Valin <Jean-Marc.Valin@xxxxxxxxxxxxxx> |
| Date: | Thu, 27 May 2004 15:05:40 -0400 |
| Cc: | fam@xxxxxxxxxxx, Nautilus mailing <nautilus-list@xxxxxxxxx> |
| In-reply-to: | <1085683907.3438.54.camel@xxxxxxxxxxxxxxxx> |
| Organization: | Universite de Sherbrooke |
| References: | <1085644229.3221.16.camel@xxxxxxxxxxxxxxxxxxxx> <1085683907.3438.54.camel@xxxxxxxxxxxxxxxx> |
| Sender: | fam-bounce@xxxxxxxxxxx |
> It would be interesting if fam is monitoring the directory you are
> working in.
> Check
> $ fuser -v /directory
> or
> $ lsof | grep /directory
Yes, I see a bunch of things there:
USER PID ACCESS COMMAND
/home/jm/phd/jmvalin/overflow
jm 3340 ..c.. zsh
jm 3346 ..c.. batchflow
jm 3347 ..c.. grep
jm 3348 ..c.. grep
> Your experience is backing my impression that file alteration monitoring
> via events (kernel directory notification using dnotify) is less
> efficient than people tend to expect...
>
> What you could try:
>
> to disable fam in xinetd
> or
> recompile fam without the dnotify patch.
Actually, the main thing is that I don't think fam should ever be
enabled by default on a system. Such unpredictable behaviour is simply
not acceptable from a daemon.
Jean-Marc
--
Jean-Marc Valin
http://www.xiph.org/~jm/
LABORIUS
Université de Sherbrooke, Québec, Canada
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: fam using lots of CPU, nf |
|---|---|
| Next by Date: | Re: fam using lots of CPU, Manuel Amador (Rudd-O) |
| Previous by Thread: | Re: fam using lots of CPU, nf |
| Next by Thread: | Re: fam using lots of CPU, Manuel Amador (Rudd-O) |
| Indexes: | [Date] [Thread] [Top] [All Lists] |