Bug 1033765

Summary: CalDAV backend doing repeated requests
Product: Red Hat Enterprise Linux 7 Reporter: Matěj Cepl <mcepl>
Component: evolution-data-serverAssignee: Matthew Barnes <mbarnes>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Desktop QE <desktop-qa-list>
Severity: high Docs Contact:
Priority: unspecified    
Version: 7.0CC: fidencio, mcrha, riehecky
Target Milestone: rc   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2015-05-22 11:03:56 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Attachments:
Description Flags
number of accesses to the webserver between 17. and 22. 11 none

Description Matěj Cepl 2013-11-22 20:09:08 UTC 
Created attachment 827993 [details]
number of accesses to the webserver between 17. and 22. 11

Description of problem:
I have discovered that my server is going inactive. After a lot of wrangling with it, I have discovered a small moment when the load average on the server was just [...] over 90, and I had over 200 httpd processes.

After digging through httpd logs I have discovered (see attached output of cut -d' ' -f 3,6- /var/log/httpd/luther.ceplovi.cz-access_log|sort|uniq -c|sort -n) that it is actually Evolution pounding on the CalDAV server there.

I have installed Evolution in the afternoon of November 20, so this

[root@luther ~]# grep 'Request from User : matej' /var/log/zarafa/ical.log|cut -d' ' -f 1-3|sort|uniq -c|sort -n -k 4
     48 Sun Nov 17
    147 Mon Nov 18
     85 Tue Nov 19
 157866 Wed Nov 20
  58004 Thu Nov 21
    642 Fri Nov 22
[root@luther ~]# 

looks like a pretty convincing proof of guilt (I have removed the calendar from Evolution this morning after the whole server melted down under OOM killer getting outraged).

Not sure, how can I analyze the issue on the client (and also I don't want to torture my poor server too much).

Version-Release number of selected component (if applicable):
evolution-3.8.5-10.el7.x86_64
zarafa-server-7.0.13-1.el6.i686
zarafa-7.0.13-1.el6.i686
zarafa-ical-7.0.13-1.el6.i686


How reproducible:
happened twice in the row (but that's two times more than I would like to)

Steps to Reproduce:
1.just add Zarafa calendar to the Evolution as a CalDAV calendar
2.
3.

Actual results:
server meltes down under the sustained DOS from Evolution

Expected results:
smoothly working shared calendar

Additional info:
Maybe there is some relationship to bug 1030579, but obviously this is in the end something different (e.g., DNS v. CalDAV).
Comment 2 Matthew Barnes 2013-12-12 15:50:39 UTC 
Too vague.  Need a backtrace of evolution-calendar-factory or better reproducer steps.
Comment 3 Matěj Cepl 2014-01-20 14:05:34 UTC 
(In reply to Matthew Barnes from comment #2)
> Too vague.  Need a backtrace of evolution-calendar-factory or better
> reproducer steps.

Sorry, I won’t try third time to kill my server (happened to me twice). I am willing to give you or Milan ssh access to the server, but it is my only home server and I (and my family) needs it for something else than doing a work for developers who wants me to do their work.

Reproducer for me is simple: enable CalDAV account in Evo. Some hours (perhaps a day or two) is hit by couple of CalDAV requests per second, eventually gets to around 100 system load and OOM killer will gradually kill whole system. No, sorry I don’t see anything wrong in Evolution, and it seems to be generally happy to kill my server. Is it really a reporter’s job to resolve the issues for you?

Closed as INSUFFICIENT_DATA if you want, but I won’t touch Evo unless I see at least some interest in fixing this other than throwing the issue on my head.
Comment 4 Milan Crha 2014-01-20 17:29:17 UTC 
I tried to reproduce this, with Matej's help, but no luck. I'll try some corner cases and will see. One observation: if I have stored a wrong password for the CalDAV calendar, the user is not asked to enter a correct one for some reason.
Comment 5 RHEL Program Management 2014-03-22 06:28:56 UTC 
This request was not resolved in time for the current release.
Red Hat invites you to ask your support representative to
propose this request, if still desired, for consideration in
the next release of Red Hat Enterprise Linux.
Comment 7 Milan Crha 2015-05-20 15:41:27 UTC 
Maybe a similar issue as here [1], except you do not use OAuth2, aka Bearer, authentication against your Zarafa server. I still think that this might be related to failed authentication and an attempt of the CalDAV backend to recover after a failed connection.

Could you run evolution-calendar-factory with CalDAV debugging on and try to reproduce the issue, then check what will be in the log, please? The command is:

   $ CALDAV_DEBUG=all /usr/libexec/evolution-calendar-factory -w &>log.txt

then kill evolution-alarm-notify and restart evolution.

[1] https://git.gnome.org/browse/evolution-data-server/commit/?id=94f01174562
Comment 8 Matěj Cepl 2015-05-22 11:03:56 UTC 
I don't have the Zarafa server anymore.