Bug 145881 – parallel-fetch runs even when files are already fetched

• Status: RESOLVED WONTFIX
• Product: Portage Development

[Description Toralf Förster 2006-09-01 10:53:02 UTC]

The message should only be printed if there is really something to fetch.

[Comment 1 Jakub Moc (RETIRED) 2006-09-01 10:56:47 UTC]

I don't see what you mean here. I don't get any message when there's nothing to fetch.

[Comment 2 Toralf Förster 2006-09-01 11:41:13 UTC]

I resumed an emerge within my user mode linux image and got the message:

n22 / # emerge --resume
*** Resuming merge...
>>> starting parallel fetching
>>> Emerging (1 of 111) sys-devel/gcc-4.1.1 to /
>>> checking ebuild checksums ;-)
>>> checking auxfile checksums ;-)
>>> checking miscfile checksums ;-)
...

Previously I ran "$> emerge -eavf world" and then "$> emerge -eav world". B/c the previously emerge command already compiled 80 of 191 packages (within some hours) I think that's enough time to fetch all other packages.

So might be that this bug does only appears for the "$> emerge --resume" option ?  BTW I do re-emerge the whole UML system by chrooting into the UML image at this way:

n22 ~ # modprobe loop && sleep 1 && mount -o loop /opt/uml/root_fs /mnt/loop && mount -t proc none /mnt/loop/proc && mount -o bind,ro /usr/portage /mnt/loop/usr/portage && mount -o bind /mnt/tmp /mnt/loop/var/tmp/portage && chroot /mnt/loop/ /bin/bash

Bug reopened, b/c it does not work for me.

[Comment 3 Jakub Moc (RETIRED) 2006-09-01 12:03:05 UTC]

(In reply to comment #2)
> n22 ~ # modprobe loop && sleep 1 && mount -o loop /opt/uml/root_fs /mnt/loop &&
> mount -t proc none /mnt/loop/proc && mount -o bind,ro /usr/portage
> /mnt/loop/usr/portage && mount -o bind /mnt/tmp /mnt/loop/var/tmp/portage &&
> chroot /mnt/loop/ /bin/bash

???

[Comment 4 Zac Medico 2006-09-01 12:23:36 UTC]

parallel-fetch is supposed to run every time.  We could optimize it so that it uses a database of successfully fetched files (with timestamps) to skip files that are known good, but even the, we'd still run parallel-fetch every time to make sure that the files exist in $DISTDIR.

[Comment 5 Toralf Förster 2006-09-01 12:40:58 UTC]

(In reply to comment #4)
Ah thanks, closing the bug.

(In reply to comment #3)
> ???
modprobe loop && sleep 1
        sleep b/c under heavy cpu load the next mount command sometimes failed
mount -o loop /opt/uml/root_fs /mnt/loop :
        the UML root file system itself (1 GB file: Gentoo base system, Apache, PHP, MySQL, Courier, NFS, some other app to play with them)
mount -t proc none /mnt/loop/proc
        needed to emerge mysql within a chrooted env 
mount -o bind,ro /usr/portage /mnt/loop/usr/portage
        use the host system, saves space and cpu
mount -o bind /mnt/tmp /mnt/loop/var/tmp/portage
        use temp space from the host -> 1 GB for root fs is enough)
chroot /mnt/loop/ /bin/bash
        enter the UML now

Ah, and in my /etc/conf.d/local.stop I have for this:

for m in $(mount | grep loop | cut -f3 -d' ' | sort -ru)
do
        umount $m
done

where the "-ru" is important.