In data domenica 2 settembre 2018 10:21:38 CEST, Bernd Nachtigall ha scritto:
> Am 31.08.2018 um 12:11 schrieb stakanov:
> > In data venerdì 31 agosto 2018 07:35:48 CEST, Bernd Nachtigall ha scritto:
> (...)
>
> >> How can I disable akonadi so that there is no restart at each new login?
>
> (...)
>
> > Depending on the version you use, the PITA may well be mariadb with
> > akonadi.
> Hi stakanow,
>
> I remember that I seems you have a hight knowledge about KDE :-)
>
> Perfect! Now somebody with a expirience in KDE development could repair
> the postgreSQL10 (or maybe better figure out how to use SQLite. So it is
> not neccessary to run a big DB-Service) issues and configure a STABLE
> and EVERYWHERE RUNABLE configuration for akonadi ;-)
> After this I think a lot of users come back to Kontact and it very good
> integration in the desktop.
>
> Because I left Kontact et. al. I want to find a solution to disable
> akonadi_davgroupware_resource. This brings regulay a lot of messages to
> give the pwd for kWallet.
>
> Just now I have added 'aconadictl stop' to a file in Autostart
> (~/.kde/Autostart/stopakonadi.sh) but do not know if this is the
> correct/recommended way.
>
> > If you do not have opensuse Leap YMMV. I can obviously not know.
>
> Indeed, I should have give this information ;-)
>
> # zypper se -i -s kded
> Repository-Daten werden geladen...
> Installierte Pakete werden gelesen...
>
> S | Name | Typ | Version | Arch | Repository
> --+-----------+-------+------------+--------+-----------------------
> i | kded | Paket | 5.32.0-1.3 | x86_64 | openSUSE-Leap-42.3-Oss
> i | kded-lang | Paket | 5.32.0-1.3 | noarch | openSUSE-Leap-42.3-Oss
>
>
> Bernd
OK, I see. With that version of KDE I had the following issues:
the indexes did corrupt and you had to run on a regular basis: close with exit
kontact, open akonadikonsole, clear the cache of the folder hit by it and then
you have to exit akonadiconsole (because the menu point forseen to restart
akonadi is crashing the program.
So you close it and run: akonadictl stop.
Then, you should run kontact/kmail, which will also restart akonadi. Open the
folder in question and wait for it to be repopulated.
Now in terminal (with kmail/kontact running run akonadictl fsck
when done, I generally tend to close all an run akonadictl restart. Maybe it
is possible to restart akonadi with kontact running, by my experience is it
gets unresponsive.
I had severe filter problems after the update. You can take every filter, make
a newone (call it filtern+n to distinguish it (so dhl gets dhl1 e.g.) so you
can copy the values from the old to the new filter. You then eliminate the old
filter and save. I am not yet sure why this is necessary (it seems a format
problem of the database version or of the way kontact sets up filters?) but it
gets you rid of the high load of CPU.
You could (with 42.3) really try to use postgressql96 (it is supported still a
long time, AFAIK more than two full year, in the meanwhile I would hope the
problems are either solved or akonadi is recognized as a problem by its very
design and abandoned.
To do so:
backup the following folders (by copying them into a new directory and
renaming them to old e.g.) so that, if everything fails you can go back. What
I tell you now is a procedure you run on your won risk so a backup of your /
home seems advisable!
How to change to postgres96 in opensuse Leap (advice and kudos go to the users
"knurpth" and "ianseeks" of the mailing list:
Install postgresql 9.6
This will install also postgres10. In order to use 9.6 you have to install the
package: yast-update-alternatives. You can do so via yast. This allows you to
select it in various and in a GUI to impose the use of the version.
--config postgresql (by upgrade alternatives)
Stop akonadi, move ~/.local/share/akonadi so that you can go back if things
go South on you. (So copy it to an extra directory and give it the "old" name.
Change ~/.config/akonadi/akonadiserverrc like this (I would advise to save the
original copy with "old" too.
Back up also "local-mail" with another directory and "old-name".
==============
[Debug]
Tracer=null
[%General]
Driver=QPSQL
[QPSQL]
Host=/tmp/akonadi-YOURUSERNAME.SOMEHASH
InitDbPath=/usr/bin/initdb
Name=akonadi
Options=
ServerPath=/usr/bin/pg_ctl
StartServer=true
==============
were yourusername is the username you have and "some hash" is some weired
combination you see when you open the original file (to be copied exactly).
Start akonadi and the database will be rebuilt. For me (with an old ironlake)
it took a day to have the whole indexing file rebuild. On new machines it may
be as short as 20 minutes. YMMV depending on the size of your hardware. Be
patient!
When having renamed the databases you can also do all in the CLI instead of
yast (whatever is better for you).
cli, as root. After stopping akonadi, and the backup of the
mysql database, local-mail, changing the akonadiserverrc, etc etc.
su
zypper in libQt5Sql5-postgresql
# or the equivalent for your Leap version
zypper in postgresql96-server
update-alternatives --config postgresql
# choose the option for 96
systemctl enable postgresql.service
systemctl start postgresql.service
exit
As user
akonadictl start
Worked for me like charm, however: control after the change the accounts and
correct the directories were the stuff should go after the change. I had a
unwanted change there (can be corrected easily, one time operation).
If you have setup you filter before O.K. If not, as said above, case of
filter problems you could then proceed to set them up anew.
For me that lowered the CPU charge and the system temperature note-worthily.
But I did that with 15, with 42.3 I changed because of the indexing problems.
What Leap 15 has is (or by akonadi or by qt-web) and enormous memory leak
after suspend to disk and waking up. There I get up to 8 GB konsumption by
akonadi alone and a 4 GB SWAP with high CPU load. If you:
exit Kontact (by killing it, the program does get unresponsive)
do akonadictl stop.
Then restart Kontact, all will be normal (e.g. 2.5 GB for three users at most)
for the rest of the day.
This bug is connected with two other bugs of 15. Baloo indexer crashes at
startup (bug known to kde) and a silent crash of Kontact if you close it and
just restart it. The first time in 15 it crashes silently. Then a restart has
normal function by you may experience some memory consumption.
This does not happen if you close Kontact and you stop akonadi. Restart
kontact all is great.
With 15 I had no LR conflicts an "unretrivable" etc errors like in 42.3. But
instead the other problems and, up to the change to postgres96 a high load of
CPU.
You may try if Postgres works for you in 42.3. I think that would be the
advisable solution up to EOL of it, which has been lengthened and would
provide you with some peace.
P.S. BACKUP! I know why I tell it, it is ALWAYS better to have it. So no tears
afterwards. ;-)
_________________________________________________________________
________________________________________________________
Ihre E-Mail-Postfächer sicher & zentral an einem Ort. Jetzt wechseln und alte E-Mail-Adresse mitnehmen! https://www.eclipso.de
