Aw: LDAP-Problems after Update Kolab 16

[Bartosiak-Jentys, Chris]

Hi all,

I have managed to consistently reproduce this issue:

kolab.conf:
modules = resources, invitationpolicy

1) send a recurring event (I choose infinitely recurring) and invite 
another user on the same kolab server.

Result wallace locks up directory server and has high CPU utilisation.

I updated to wallace-0.8.3-3.1.el7.kolab_16.noarch, same issue there as 
on wallace-0.8.1-3.3.el7.kolab_16.noarch.

As soon as I pkill -9 wallaced dirsrv jumps back to life and kolab 
(webmail etc.) starts working again.

If I remove invitationpolicy from modules in kolab.conf:
modules = resources, invitationpolicy
to
modules = invitationpolicy

I can send invites to recurring events without issue.


I will try and gather some logs and file a bug report.

Chris.


On 2016-07-25 10:21, Bartosiak-Jentys, Chris wrote:
> Hi All,
> 
> Roland & guldendraak, thank you for posting to the list! I ran into
> the same issue recently and spent days looking for issues in dirsrv.
> The symptoms I experienced were identical to those you experienced.
> 
> In the end I created a new directory instance from scratch and it now
> seems stable for a week now though I haven't been testing calendar
> invites which may well trigger the issue as wallaced is largely
> responsible for automatically handling calendar invites (as well as
> other functions like inserting footers etc.).
> 
> My kolab.conf settings for wallace are:
> modules = resources, invitationpolicy
> 
> Though I can't recall if these were the settings in place when I
> encountered the bug as I was experimenting with automatic invitation
> handling. I also had custom invitation handling policies in place for
> one user configured through webadmin which I guess invitationpolicy in
> kolab.conf
> When I encountered the issue I was running 
> 0.8.1-2.4.el7.kolab_16.noarch.
> 
> I'm now on wallace-0.8.1-3.3.el7.kolab_16.noarch I will try and
> trigger the problem again and test the wallace settings in kolab.conf
> you listed.
> 
> I will post back when I have done some testing, I have a feeling that
> calendar invites triggered this.
> 
> Thanks, looking for issues in my directory server was driving me 
> insane!
> 
> Chris
> 
> 
> On 2016-07-24 14:22, guldendraak at gmx.net wrote:
>> Hi Roland
>> 
>> this sounds really familiar. Yesterday I spent about 5 hours
>> troubleshooting a similar problem. Here is what I did, maybe it is
>> helpful to you as well, but first some info about my setup:
>> 
>> - Kolab 16 on CentOS, latest package versions (some were updated on
>> the 22nd this month)
>> - multi domain installation as per instructions from here:
>> https://docs.kolab.org/howtos/multi-domain.html
>> 
>> It turns out that wallace has a problem with certain combinations of
>> the modules. This is what I found (defined in /etc/kolab/kolab.conf):
>> 
>> ;modules = resources, invitationpolicy, footer  # NOK!
>> ;modules = invitationpolicy                     # OK
>> ;modules = resources                            # OK
>> ;modules = footer                               # OK
>> ;modules = invitationpolicy, resources          # NOK!
>> ;modules = invitationpolicy, footer             # OK
>> ;modules = resources, footer                    # OK
>> 
>> When using one of the settings marked with NOK, I get these symptoms:
>> 
>> - postmap -q ldap:/<path-to-file> completely fails (timeout)
>> - Users are no longer able to log in
>> - Wallace and Kolab randomly use 100% CPU
>> - systemctl restart wallace takes forever (like several minutes), but
>> as soon as this command is issued, logons are possible again, even
>> before the command completes
>> 
>> When Wallace uses 100% CPU, an strace attached to the process shows
>> extremely rapid polling attempts to one of the file descriptors opened
>> by dirsrv. This seems to make dirsrv completely unresponsive also for
>> other clients, therefore logins and postmap don't work while this is
>> happening. I spent the good part of my time trying to troubleshoot
>> dirsrv only to find out that it doesn't have an issue at all.
>> 
>> As a side note: this only started happening _after_ I upgraded some
>> kolab packages on the 22nd (if I remember correctly, some packages
>> were upgraded from 0.8.1 to 0.8.3). Before, Wallace would simply do
>> nothing regarding the invitation policy, but at least it wasn't
>> killing the system. I am wondering what happens if we just not use
>> wallace at all. Do you know the complete list of things it is supposed
>> to do in Kolab?
>> 
>> Cheers
>> 
>>> Gesendet: Sonntag, 24. Juli 2016 um 14:26 Uhr
>>> Von: "Roland Kolb" <roland.kolb at ib-ulherr.de>
>>> An: users at lists.kolab.org
>>> Betreff: LDAP-Problems after Update Kolab 16
>>> 
>>> Hi all,
>>> 
>>> after the update of Kolab 16 (Centos7) I get problems connecting to 
>>> Kolab.
>>> 
>>> A few times per day the connection to the ldap server will be lost. A
>>> login to a mail account isn't possible. After about 30min the 
>>> connection
>>> to the ldap server is possible again and also a login the the mail 
>>> account.
>>> When I lool into the log files I found
>>> 
>>> kolab postfix/trivial-rewrite[21915]: warning: dict_ldap_lookup: 
>>> Search
>>> error -5: Timed out
>>> kolab postfix/trivial-rewrite[21915]: warning:
>>> ldap:/etc/postfix/ldap/mydestination.cf: table lookup problem
>>> 
>>> Has somebody an idea what happens why the connection will be lost? My
>>> ldap-server
>>> 
>>> 389-ds-base.x86_64           (1.3.4.0-32.el7_2)
>>> 389-ds-base-libs.x86_64    (1.3.4.0-32.el7_2)
>>> 
>>> Thanks
>>> 
>>> Roland
>>> _______________________________________________
>>> users mailing list
>>> users at lists.kolab.org
>>> https://lists.kolab.org/mailman/listinfo/users
>>> 
>> _______________________________________________
>> users mailing list
>> users at lists.kolab.org
>> https://lists.kolab.org/mailman/listinfo/users
> _______________________________________________
> users mailing list
> users at lists.kolab.org
> https://lists.kolab.org/mailman/listinfo/users