Accidentally I noticed that 'fam' disappeared (likely) from all mirrors. Last time I looked there was something like http://download.fedora.redhat.com/pub/fedora/linux/core/development/SRPMS/fa... but if you would try to retrieve that you would get "404: Page not found" instead. No recompiled binaries in "development".
At least the current nautilus version depends on fam so, unless some deep nautilus rewrite is coming, that vanishing act seem to be a result of some hiccup.
Michal
Tim Waugh wrote:
On Thu, Jul 29, 2004 at 10:10:43AM -0600, Michal Jaegermann wrote:
At least the current nautilus version depends on fam so, unless some deep nautilus rewrite is coming, that vanishing act seem to be a result of some hiccup.
Install gamin, fam's replacement.
Tim. */
After my 7/23/04 upgrade from rawhide, gamin was pulled in. There were some problems with some error messages though.
Upgrading gamin-0.0.2-1.i386. warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... Upgrading gamin-devel-0.0.2-1.i386.
I guess this is selinux related.
Jim
On Thu, Jul 29, 2004 at 08:56:41PM -0400, Jim Cornette wrote:
After my 7/23/04 upgrade from rawhide, gamin was pulled in. There were some problems with some error messages though.
Upgrading gamin-0.0.2-1.i386. warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... Upgrading gamin-devel-0.0.2-1.i386.
I guess this is selinux related.
yes, but I can't really understand what's it's reporting,
Daniel
On Thu, 2004-07-29 at 20:56, Jim Cornette wrote:
After my 7/23/04 upgrade from rawhide, gamin was pulled in. There were some problems with some error messages though.
Upgrading gamin-0.0.2-1.i386. warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... Upgrading gamin-devel-0.0.2-1.i386.
I guess this is selinux related.
What version of policy are you running? Latest is selinux-policy-{strict,targeted}-1.15.10-1, I think.
Stephen Smalley wrote:
On Thu, 2004-07-29 at 20:56, Jim Cornette wrote:
After my 7/23/04 upgrade from rawhide, gamin was pulled in. There were some problems with some error messages though.
Upgrading gamin-0.0.2-1.i386. warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... warning: setexeccon(user_u:system_r:rpm_script_t) fails from context "user_u:system_r:unconfined_t": Invalid argument Continuing ... Upgrading gamin-devel-0.0.2-1.i386.
I guess this is selinux related.
What version of policy are you running? Latest is selinux-policy-{strict,targeted}-1.15.10-1, I think.
The update occured on 7/23 and the mirror was ncsu. I believe this mirror lags others by a day. The rawhide packages were updated just before performing the upgrade.
From the upgrade log. here are the versions upgraded.
Upgrading selinux-policy-strict-1.15.7-4.noarch. matchpathcon(/etc/selinux/config) failed Invalid argument Upgrading selinux-policy-targeted-1.15.7-4.noarch.
Errors on the install related to SELinux were posted in an earlier mailing to the list below.
http://www.redhat.com/archives/fedora-test-list/2004-July/msg00542.html
On Fri, 2004-07-30 at 18:50, Jim Cornette wrote:
The update occured on 7/23 and the mirror was ncsu. I believe this mirror lags others by a day. The rawhide packages were updated just before performing the upgrade.
From the upgrade log. here are the versions upgraded.
Upgrading selinux-policy-strict-1.15.7-4.noarch. matchpathcon(/etc/selinux/config) failed Invalid argument Upgrading selinux-policy-targeted-1.15.7-4.noarch.
Errors on the install related to SELinux were posted in an earlier mailing to the list below.
http://www.redhat.com/archives/fedora-test-list/2004-July/msg00542.html
Hmm..rpm_script_t is an alias for unconfined_t in the targeted policy.
$ id -Z user_u:system_r:unconfined_t $ runcon -t rpm_script_t bash $ id -Z user_u:system_r:unconfined_t $ rpm -q selinux-policy-targeted selinux-policy-targeted-1.15.10-1
Michal Jaegermann michal@harddata.com writes:
Accidentally I noticed that 'fam' disappeared (likely) from all mirrors. Last time I looked there was something like http://download.fedora.redhat.com/pub/fedora/linux/core/development%5C /SRPMS/fam-2.6.10-11.src.rpm but if you would try to retrieve that you would get "404: Page not found" instead. No recompiled binaries in "development".
Michal,
I butted in here on the hunch that you or some one else might be able to advise me about `fam'. I've never knowingly used `fam' but it looks like something that could report file changes.
Only trouble is the docu only says it reports file changes to other apps. Never mentions how one might use it to get a report in a file.
Can `fam' do such a thing? If so can you steer me to some howto info that will help me set it up?
I'm interested in logging file changes in a given directory. It consisted some 2-3 thousand text files.
I'd hoped to script up something using fam to do the checking.
On Thu, 2004-07-29 at 20:50 -0500, Harry Putnam wrote:
Michal Jaegermann michal@harddata.com writes:
Accidentally I noticed that 'fam' disappeared (likely) from all mirrors. Last time I looked there was something like http://download.fedora.redhat.com/pub/fedora/linux/core/development%5C /SRPMS/fam-2.6.10-11.src.rpm but if you would try to retrieve that you would get "404: Page not found" instead. No recompiled binaries in "development".
Michal,
I butted in here on the hunch that you or some one else might be able to advise me about `fam'. I've never knowingly used `fam' but it looks like something that could report file changes.
Only trouble is the docu only says it reports file changes to other apps. Never mentions how one might use it to get a report in a file.
Can `fam' do such a thing? If so can you steer me to some howto info that will help me set it up?
I'm interested in logging file changes in a given directory. It consisted some 2-3 thousand text files.
I'd hoped to script up something using fam to do the checking.
Morning Michal
I believe fam primarily looks for new files in directories, updates in file details on the file system, e.g. access time, permissions etc. It then lets programs know.
It's more monitoring changes in the file system then changes in the files them selves (or at least that is my impression).
I get the feeling your looking more for some thing along the lines of diff, which compares new and old files. Does that sound correct?
On Fri, Jul 30, 2004 at 08:11:56AM +0100, Douglas Furlong wrote:
On Thu, 2004-07-29 at 20:50 -0500, Harry Putnam wrote:
Michal,
I butted in here on the hunch that you or some one else might be able to advise me about `fam'. I've never knowingly used `fam' but it looks like something that could report file changes.
Only trouble is the docu only says it reports file changes to other apps. Never mentions how one might use it to get a report in a file.
Can `fam' do such a thing? If so can you steer me to some howto info that will help me set it up?
I'm interested in logging file changes in a given directory. It consisted some 2-3 thousand text files.
I'd hoped to script up something using fam to do the checking.
Morning Michal
I believe fam primarily looks for new files in directories, updates in file details on the file system, e.g. access time, permissions etc. It then lets programs know.
It's more monitoring changes in the file system then changes in the files them selves (or at least that is my impression).
I get the feeling your looking more for some thing along the lines of diff, which compares new and old files. Does that sound correct?
The FAM API is about reporting changes, but not how files changed. It is relatively expensive on large directories unfortunately: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=124361
I will try to reduce the CPU usage in gamin but we lack kernel support which would make that easy.
Daniel
On Fri, 30 Jul 2004 03:35:26 -0400, Daniel Veillard veillard@redhat.com wrote:
The FAM API is about reporting changes, but not how files changed. It is relatively expensive on large directories unfortunately: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=124361
I will try to reduce the CPU usage in gamin but we lack kernel support which would make that easy.
On a related note, is there a way to prevent FAM from locking up writable removable media? Sometimes - but not always - when using FAM in conjunction with Nautilus, I get unmount errors and have had to killall -9 fam to proceed.
Regards,
- Michel
On Fri, Jul 30, 2004 at 07:00:49PM +0700, Michel Salim wrote:
On Fri, 30 Jul 2004 03:35:26 -0400, Daniel Veillard veillard@redhat.com wrote:
The FAM API is about reporting changes, but not how files changed. It is relatively expensive on large directories unfortunately: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=124361
I will try to reduce the CPU usage in gamin but we lack kernel support which would make that easy.
On a related note, is there a way to prevent FAM from locking up writable removable media? Sometimes - but not always - when using FAM in conjunction with Nautilus, I get unmount errors and have had to killall -9 fam to proceed.
Known problem. This is being worked on at various level. I can't promise that gamin will be able to do that, because as long as the dnotify API is used one need to open the directory/file being monitored. Yes this sucks, and unfortunately I don't expect to see any kernel API improvement at this point.
Daniel
On Fri, 30 Jul 2004 08:05:42 -0400, Daniel Veillard veillard@redhat.com wrote:
Known problem. This is being worked on at various level. I can't promise that gamin will be able to do that, because as long as the dnotify API is used one need to open the directory/file being monitored. Yes this sucks, and unfortunately I don't expect to see any kernel API improvement at this point.
Just filed a bug on this 128905 ...but its ontopic for this thread. I ran into this with gamin just now on my test box... gamin is holding the mount points like fam did. But there is a twist... with fam i was able to kill the fam process and then unmount with no obvious harm done. I try to do that with gam_server.. and gnome panel freaks out, and appears to die..i get a whole slew of messages about the panel restarting and applets reloading. And gam_server is back with a new process number and I still cant unmount.
I finally tricked gamin to unmount with this commandline: kill -9 15365; umount /mnt/flash ; umount /mnt/flash1
man oh man.. gnome panel really really didnt like that. Not only did i get the standard pop-up windows about panel restarting... but i got half a dozen or so empty dialog windows...empty of all text.. no buttons no text. I didnt even see the panel restart dialog until i closed all the blank windows. But this did unmount the flash mountpoints. Hopefully my experience trying to convince gam_server to let go of the mounts wasn't a design goal. Though i did experience a profound moment of contemplation as i stared at the dialogless dialog boxes. As a modern art piece...it was breath taking.
-jef"if i only knew what i could bribe gam_server with, i'd barter with it to give me back my usb devices"spaleta
On Sat, Jul 31, 2004 at 12:16:36AM -0400, Jeff Spaleta wrote:
On Fri, 30 Jul 2004 08:05:42 -0400, Daniel Veillard veillard@redhat.com wrote:
Known problem. This is being worked on at various level. I can't promise that gamin will be able to do that, because as long as the dnotify API is used one need to open the directory/file being monitored. Yes this sucks, and unfortunately I don't expect to see any kernel API improvement at this point.
Just filed a bug on this 128905 ...but its ontopic for this thread. I ran into this with gamin just now on my test box... gamin is holding the mount points like fam did. But there is a twist... with fam i was able to kill the fam process and then unmount with no obvious harm done. I try to do that with gam_server.. and gnome panel freaks out, and appears to die..i get a whole slew of messages about the panel restarting and applets reloading. And gam_server is back with a new process number and I still cant unmount.
I finally tricked gamin to unmount with this commandline: kill -9 15365; umount /mnt/flash ; umount /mnt/flash1
man oh man.. gnome panel really really didnt like that. Not only did i get the standard pop-up windows about panel restarting... but i got half a dozen or so empty dialog windows...empty of all text.. no buttons no text. I didnt even see the panel restart dialog until i closed all the blank windows. But this did unmount the flash mountpoints. Hopefully my experience trying to convince gam_server to let go of the mounts wasn't a design goal. Though i did experience a profound moment of contemplation as i stared at the dialogless dialog boxes. As a modern art piece...it was breath taking.
I pushed fam-0.0.3 onto rawhide today. Could you try it out, rebuild from src.rpm at: http://www.gnome.org/~veillard/gamin/
This may help a bit, or not ... :-)
Daniel
Douglas Furlong douglas.furlong@firebox.com writes:
It's more monitoring changes in the file system then changes in the files them selves (or at least that is my impression).
I get the feeling your looking more for some thing along the lines of diff, which compares new and old files. Does that sound correct?
Yes. I'm thinking of something that reports content of file changes as they happen. This something might employ `diff' I guess. Maybe a homemade daemon that notices changes then runs diff...