akonadi eats CPU after suspend

classic Classic list List threaded Threaded
17 messages Options
Reply | Threaded
Open this post in threaded view
|

akonadi eats CPU after suspend

Francesco Montesano
Hi,

On my TW machine, since a couple of weeks, akonadi behaves badly after
resuming from suspend.

Most of the times I have two akonadi instances using each one about
one CPU, plus there is some mysql instance, related to akonadi, that
eats some extra 30 or 40% of cpus.

This slows down the computer very much. Usually "akonadictl restart"
fixes the problem.

I've looked on google for some more info, but I haven't found anything.

I use kmail with only me work emain via imaps
Plus I did connect my google address book to KAdressBook (I would like
to remove it, but I can't find where)
If relevant, I use telepathy with Hangout.

Is there a place with some log I can check or something like this or a
workaround?

Ciao,

Francesco
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Luca Beltrame
In data martedì 26 aprile 2016 17:04:27 CEST, Francesco Montesano ha scritto:

Hello,

> On my TW machine, since a couple of weeks, akonadi behaves badly after
> resuming from suspend.

Can you post your PIM stack versions? I haven't seen this in a while on my
laptop.

--
Luca Beltrame - KDE Forums team
KDE Science supporter
GPG key ID: A29D259B

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

2016-04-26 17:07 GMT+02:00 Luca Beltrame <[hidden email]>:
>
> Can you post your PIM stack versions? I haven't seen this in a while on my
> laptop.

I hope this is what you want:

kdepim: 15.12.3-2.1
kdepim-runtime: 15.12.3-2.1
kdepimlibs4: 4.14.10-1.2
kio-pimlibs: 15.12.3-1.1
libKF5PimTextEdit5: 15.12.3-1.1
libbaloopim4: 4.14.3-2.2
libkdepim: 15.12.3-2.1
libkdepimlibs4: 4.14.10-1.2
kmailtransport: 15.12.3-1.1
kmail5: version 15.12.3-2.1

The Kmail application:
kmail2 5.1.3
Using:
KDE Frameworks 5.21.0
Qt 5.5.1 (built against 5.5.1)
The xcb windowing system

They all come from repo-oss.
I also use kde-extra repo.

Ciao,

Francesco

>
> --
> Luca Beltrame - KDE Forums team
> KDE Science supporter
> GPG key ID: A29D259B
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

auxsvr
In reply to this post by Francesco Montesano
On Tuesday, 26 April 2016 17:04:27 EEST Francesco Montesano wrote:
> Hi,

Hi,
 
> On my TW machine, since a couple of weeks, akonadi behaves badly after
> resuming from suspend.
>
> Most of the times I have two akonadi instances using each one about
> one CPU, plus there is some mysql instance, related to akonadi, that
> eats some extra 30 or 40% of cpus.

> I use kmail with only me work emain via imaps
> Plus I did connect my google address book to KAdressBook (I would like
> to remove it, but I can't find where)

Akonadi is known to loop forever updating resources such as remote .ics
files from Google, if you set them read-only. Perhaps this is a similar
problem in your case? You may remove the address book from the left
column in KAddressBook or by using akonadiconsole.

> Ciao,
>
> Francesco
--
Regards,
Peter
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

2016-04-26 18:00 GMT+02:00 auxsvr <[hidden email]>:

> On Tuesday, 26 April 2016 17:04:27 EEST Francesco Montesano wrote:
>> Hi,
>
> Hi,
>
>> On my TW machine, since a couple of weeks, akonadi behaves badly after
>> resuming from suspend.
>>
>> Most of the times I have two akonadi instances using each one about
>> one CPU, plus there is some mysql instance, related to akonadi, that
>> eats some extra 30 or 40% of cpus.
>
>> I use kmail with only me work emain via imaps
>> Plus I did connect my google address book to KAdressBook (I would like
>> to remove it, but I can't find where)
>
> Akonadi is known to loop forever updating resources such as remote .ics
> files from Google, if you set them read-only. Perhaps this is a similar
> problem in your case? You may remove the address book from the left
> column in KAddressBook or by using akonadiconsole.

I had my google calendar and I removed it, so it's not that.
About removing the address book: if I right click on
my account entry, I see a "delete address book".
Does it only deletes the address book from the akonadi list
or also tells google about deleting stuff?

Fra

>
>> Ciao,
>>
>> Francesco
> --
> Regards,
> Peter
> --
> To unsubscribe, e-mail: [hidden email]
> To contact the owner, e-mail: [hidden email]
>
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

2016-04-26 18:08 GMT+02:00 Francesco Montesano <[hidden email]>:
>>> On my TW machine, since a couple of weeks, akonadi behaves badly after
>>> resuming from suspend.
>>>

It did happen again. This time I've looked into xsession-error and journalctl.
While akonadi occupies the cpu xsession-error get the following string
many times a second:

posting retrieval request for item 28933  there are  1  queues and  0
items in mine
request for item 28933 still pending - waiting
processing retrieval request for item 28933  parts: ("RFC822")  of
resource: "akonadi_akonotes_resource_0"
akonadiagentbase_log: Item does not provide part "RFC822"
continuing
request for item 28933 succeeded

In journalctl I can't see anything about akonadi. Just awake messages
from various units, some complaints and error from telepathy and pam.
Attached you can see the first few minutes after awaking.

Ciao,

Fra

after_resume_akonadi_journalctl.log (81K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

Little update
2016-04-27 17:15 GMT+02:00 Francesco Montesano <[hidden email]>:

> Hi,
>
> 2016-04-26 18:08 GMT+02:00 Francesco Montesano <[hidden email]>:
>>>> On my TW machine, since a couple of weeks, akonadi behaves badly after
>>>> resuming from suspend.
>>>>
>
> It did happen again. This time I've looked into xsession-error and journalctl.
> While akonadi occupies the cpu xsession-error get the following string
> many times a second:
>
> posting retrieval request for item 28933  there are  1  queues and  0
> items in mine
> request for item 28933 still pending - waiting
> processing retrieval request for item 28933  parts: ("RFC822")  of
> resource: "akonadi_akonotes_resource_0"
> akonadiagentbase_log: Item does not provide part "RFC822"
> continuing
> request for item 28933 succeeded

I get this also when akonadi is not sucking all the CPU

Fra
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Luca Beltrame
In data mercoledì 27 aprile 2016 17:18:41 CEST, Francesco Montesano ha
scritto:

> I get this also when akonadi is not sucking all the CPU

It's likely unrelated then. Installing debug symbols and providing a backtrace
to the 100% CPU process would be helpful in finding the issue.

But before that, does this also occur on a network loss?

--
Luca Beltrame - KDE Forums team
KDE Science supporter
GPG key ID: A29D259B

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

2016-04-27 17:24 GMT+02:00 Luca Beltrame <[hidden email]>:
> In data mercoledì 27 aprile 2016 17:18:41 CEST, Francesco Montesano ha
> scritto:
>
>> I get this also when akonadi is not sucking all the CPU
>
> It's likely unrelated then. Installing debug symbols and providing a backtrace
> to the 100% CPU process would be helpful in finding the issue.

Whould they be the -debuginfo version of the packages I've listed
yesterday in my mail?
How would I provide a backtrace in that case?

> But before that, does this also occur on a network loss?

Not that I've noticed. Also it doesn't happen after every resume.

Ciao,

Fra

> --
> Luca Beltrame - KDE Forums team
> KDE Science supporter
> GPG key ID: A29D259B
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Martin Konold-2
Am Mittwoch, 27. April 2016, 17:33:46 CEST schrieb Francesco Montesano:

Hi,

> Whould they be the -debuginfo version of the packages I've listed
> yesterday in my mail?
> How would I provide a backtrace in that case?

You attach gdb to the running process and enter bt on the commandline.

Something like:

Use top in konsole in order to identify the cpu consuming process and its PID
(process id).

Use gdb in konsole and attach to the running process:

km82rt@sony-01:~> gdb -p 3232
GNU gdb (GDB; %maintenance_distribution) 7.9.1
Copyright (C) 2015 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-suse-linux".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://bugs.opensuse.org/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word".
Attaching to process 3232
[...]

(gdb) bt
#0  0x00007f6a759b3bbd in poll () at /lib64/libc.so.6
#1  0x00007f6a73d49e64 in  () at /usr/lib64/libglib-2.0.so.0
#2  0x00007f6a73d49f7c in g_main_context_iteration () at /usr/lib64/
libglib-2.0.so.0
#3  0x00007f6a764f7d6c in
QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) ()
at /usr/lib64/libQt5Core.so.5
#4  0x00007f6a7649ed53 in
QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () at /usr/lib64/
libQt5Core.so.5
#5  0x00007f6a764a68f6 in QCoreApplication::exec() () at /usr/lib64/
libQt5Core.so.5
#6  0x000000000047b3b2 in main ()
(gdb)

The output from the 'bt' command is relevant.

Regards
--martin konold

--
Dipl.-Physiker Martin Konold

e r f r a k o n
Erlewein, Frank, Konold & Partner - Beratende Ingenieure und Physiker
Registergericht: Amtsgericht Stuttgart PR 126
Firmensitz: Adolfstraße 23, 70469 Stuttgart
fon: 0711 67400963
fax: 0711 67400959
email: [hidden email]
http://www.erfrakon.de


--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Martin Koller
In reply to this post by Luca Beltrame
On Mittwoch, 27. April 2016 17:24:04 CEST Luca Beltrame wrote:
> In data mercoledì 27 aprile 2016 17:18:41 CEST, Francesco Montesano ha
> scritto:
>
> > I get this also when akonadi is not sucking all the CPU
>
> It's likely unrelated then. Installing debug symbols and providing a backtrace
> to the 100% CPU process would be helpful in finding the issue.

I had the same problem. backtrace attached.
there are more than 80 threads, and it seems all but thread 34 are waiting in poll().
Thread 34 is in Akonadi::Server::Connection::readCommand()

> But before that, does this also occur on a network loss?

no. Just with disconnecting or reconnecting the network it did not happen.


--
Best regards/Schöne Grüße

Martin
A: Because it breaks the logical sequence of discussion
Q: Why is top posting bad?

()  ascii ribbon campaign - against html e-mail
/\                        - against proprietary attachments

Geschenkideen, Accessoires, Seifen, Kulinarisches: www.lillehus.at

akonadi-100_cpu.txt (68K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

> On Mittwoch, 27. April 2016 17:24:04 CEST Luca Beltrame wrote:
>>
>> It's likely unrelated then. Installing debug symbols and providing a backtrace
>> to the 100% CPU process would be helpful in finding the issue.

I finally managed to get the bad guys. Attached are the tracebacks of
the two akonadiserver processes using about 1 CPU each

>> But before that, does this also occur on a network loss?

No. However it seems that: after a reboot everything works fine for
some suspend-resume cycle. Then something happens and akonadi begins
miss-behaving after at every resume.

Let me know if you need more info and/or if I should file a bug at kde

Ciao,

Francesco

akonadi-8271.txt (6K) Download Attachment
akonadi-8089.txt (113K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Luca Beltrame
In data martedì 3 maggio 2016 00:22:40 CEST, Francesco Montesano ha scritto:

> I finally managed to get the bad guys. Attached are the tracebacks of
> the two akonadiserver processes using about 1 CPU each

Thanks, those may be helpful in finding the root cause. Now you see you mention
*two* processes for "akonadiserver". Normally there should be just one and no
more, IIRC. What happens if you kill either one?

--
Luca Beltrame - KDE Forums team
GPG key ID: A29D259B

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi Luca,

2016-05-03 7:20 GMT+02:00 Luca Beltrame <[hidden email]>:
> In data martedì 3 maggio 2016 00:22:40 CEST, Francesco Montesano ha scritto:
>
>> I finally managed to get the bad guys. Attached are the tracebacks of
>> the two akonadiserver processes using about 1 CPU each
>
> Thanks, those may be helpful in finding the root cause. Now you see you mention
> *two* processes for "akonadiserver". Normally there should be just one and no
> more, IIRC. What happens if you kill either one?

When it locks I see two processes in htop using about 1 cpu each. If I
filter akonadiserver
I see some 20 instances (see attachment). Apparently htop shows all the threads,
so I'm not sure if the two on top are two processes or two threads of
the same process.
Next time I see the issue I'll investigate and let you know.

Ciao,

Fra

>
> --
> Luca Beltrame - KDE Forums team
> GPG key ID: A29D259B

Screenshot_20160503_085531.png (211K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,


2016-05-06 10:10 GMT+02:00 Francesco Montesano <[hidden email]>:
> [...]
>> Thanks, those may be helpful in finding the root cause. Now you see you mention
>> *two* processes for "akonadiserver". Normally there should be just one and no
>> more, IIRC. What happens if you kill either one?

it happened again and checked.

* ps and top show only one akonadiserver process,
* htop shows two very demanding processes as in the screenshot
attached to my previous mail,
  one of which is the one that I also see in ps/top

So I guess that htop shows also child threads.

Of the two processes in htop showing a lot of activity, I've killed
the one that is *not* shown by ps/top. The main akonadiserver process
went back to normal.

Ciao,

Fra
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Luca Beltrame
In data venerdì 6 maggio 2016 20:48:25 CEST, Francesco Montesano ha scritto:

> Of the two processes in htop showing a lot of activity, I've killed
> the one that is *not* shown by ps/top. The main akonadiserver process
> went back to normal.

Is also the system working normally after this? I think a second process is
spawned after suspend, but I've never ever seen this behavior myself.

--
Luca Beltrame - KDE Forums team
GPG key ID: A29D259B

signature.asc (465 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: akonadi eats CPU after suspend

Francesco Montesano
Hi,

2016-05-08 18:54 GMT+02:00 Luca Beltrame <[hidden email]>:
> In data venerdì 6 maggio 2016 20:48:25 CEST, Francesco Montesano ha scritto:
>
>> Of the two processes in htop showing a lot of activity, I've killed
>> the one that is *not* shown by ps/top. The main akonadiserver process
>> went back to normal.
>
> Is also the system working normally after this? I think a second process is
> spawned after suspend, but I've never ever seen this behavior myself.

Kmail works fine afterwards.

However it's not the only akonadiserver is not the only thing that
locks after suspend.
Plasmashell locks quite frequently, but before reporting I need to
check if this happens
only after disconnecting external screens (I haven't reported this
since I know that
multiscreen at the moment doesn't work properly)
Yesterday I got krunner locked because of baloo. If it happens again
I'll report at bugs.kde

Ciao,

Fra

>
> --
> Luca Beltrame - KDE Forums team
> GPG key ID: A29D259B
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]