Muss man wissen.

Dieser Inhalt wird in Deutschland nicht angezeigt. Weitere Informationen

When it comes to web-browsing, I think I would call myself a "power-user". My current feed-list has 585 entries (some of them are comment feeds, but still a large number), and the usual way of reading them is letting newsbeuter manage them, then "pre-filter" them (most feeds do not have content embedded, and newsbeuter is a console ui, which is nice but not really suitable for webcomics, etc.) and putting them on a list that I then select in Firefox. Then I will use a feature of Tab Mix Plus, namely being able to open all selected links in new tabs. Furthermore, I configured Tab Mix Plus such that it shows these tabs in multiple lines. For a long period of time, I configured it to show the tab bar at the bottom rather than the top, but this feature someday began not to work properly anymore, and I did not have time to investigate further.

At present, Firefox seems to be the only browser that can really handle many tabs properly. Chromium has some extensions that sort-of work, but it is that kind of software that "knows best". Opera had this feature in the past, too, but they dropped it. Vivaldi seems nice with respect to that. I also used tab groups, which is a feature that will soon be dropped. Now I got used to bookmarks for the same purpose.

Firefox has a good crash recovery mechanism that usually works, but it occured to me several times now that my profile got broken and I lost many of my bookmarks or at least had a lot of work restoring them, and all the settings. In the past I used Weave, which is now called Sync. However, I wanted to run my own server, because the official Mozilla server had a 5 MiB boundary, which is not enough for me (and besides that, I have a server, so I want to use it), but someday they changed the protocol and dropped support for the old protocol, and there was no easy way of setting up a new server.

These are several reasons why I am trying to keep my local Firefox profile working. So here is what I am doing:

Firstly, my Linux runs on a LVM. So I can create an additional logical volume "firefox", which I made a 512 MiB ext3 partition. This partition will be mounted (via fstab) to ~/.mozilla. And then, I use a straightforward backup procedure:

#!/usr/bin/env bash


while test -e "${SECPATH}/backup.$i"; do ((i++)); done

lvcreate -l 100%FREE --snapshot /dev/lvm1/firefox -n firefox-b || (wall "konnte kein backup erstellen!"; exit 1)
TMPFL=$(mktemp -d)
mount -o ro /dev/lvm1/firefox-b ${TMPFL} || (wall "konnte backup erstellen aber nicht mounten. tmp-directory $TMPFL wurde nicht geloescht."; exit 1)

rsync -alH --delete "${TMPFL}/" "${SECPATH}/backup.$i" --link-dest "${SECPATH}/backup.$((i-1))"

umount /dev/lvm1/firefox-b || (wall "Konnte backup /dev/lvm1/firefox-b (auf $TMPFL) nicht unmounten!"; exit 1)
lvremove -f /dev/lvm1/firefox-b || (wall "konnte /dev/lvm1/firefox-b nicht loeschen!"; exit 1)
rm -rf ${TMPFL} || (wall "konnte $TMPFL nicht loeschen!"; exit 1)

if [ $i -ge "$BACKUPNUM" ]; then
rm -rf "${SECPATH}/backup.0"
j=0; while [ $j -lt $i ]; do
mv "${SECPATH}/backup.$((j+1))" "${SECPATH}/backup.$j"

exit 0

Of course, I could just rsync the original directory, but that already broke some of my configurations (probably because it is not "atomic"). Creating an LVM snapshot leaves a directory that looks like the system has crashed – a situation Firefox can usually handle. I am running this script every 10 minutes in a loop. It is not perfect, but it serves its purpose.

Update: I now use BTRFS which has an internal snapshot mechanism. The code now looks like

while true;
do btrfs subvolume snapshot -r .mozilla/ snap/dot_mozilla_`date -u +%s`;
   sleep 600;