Bug 1069678

Summary: named Service Fails: unable to signal parent that we otherwise started successfully
Product: [Fedora] Fedora Reporter: emoziko <emobuxuti>
Component: bindAssignee: Tomáš Hozza <thozza>
Status: CLOSED NOTABUG QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 20CC: emobuxuti, psimerda, thozza, vonsch
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2014-02-27 06:50:43 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:
Embargoed:
Attachments:
Description Flags
named.run file none

Description emoziko 2014-02-25 13:59:17 UTC 
Description of problem:

I have BIND resolver installed On my Fedora 20.

I am using this server to block some bad websites, like malware or 18+, I am just redirecting this websites on 127.0.0.1 and users cannot access it.
So i have more than 500K website list to block. 

When I try to start named service with this list, it fails, but without it, it runs OK.

Also, when i try to start bind with simple command (#named -u named) it starts ok, but not with systemctl start named.service;


Version-Release number of selected component (if applicable):

[root@master ~]# uname -a
Linux master 3.13.3-201.fc20.x86_64 #1 SMP Fri Feb 14 19:08:32 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux

[root@master ~]# rpm -qa | grep bind
bind-license-9.9.4-11.P2.fc20.noarch
bind-9.9.4-11.P2.fc20.x86_64
bind-libs-lite-9.9.4-11.P2.fc20.x86_64
bind-libs-9.9.4-11.P2.fc20.x86_64
bind-utils-9.9.4-11.P2.fc20.x86_64

[root@master ~]# getenforce
Permissive


How reproducible:


Steps to Reproduce:
1. attach 500k list of zones in named.conf or in separate file and include it in named.conf
2. systemctl start named.service
3. systemctl status named.service

Actual results:
last lines of status command:

Feb 25 17:23:42 master systemd[1]: named.service operation timed out. Terminating.
Feb 25 17:23:42 master systemd[1]: Failed to start Berkeley Internet Name Domain (DNS).
Feb 25 17:23:42 master systemd[1]: Unit named.service entered failed state.

last lines of general log of bind:

25-Feb-2014 17:23:41.633 all zones loaded
25-Feb-2014 17:23:42.999 client.c:2356: unexpected error:
25-Feb-2014 17:23:42.999 isc_socket_accept() failed: shutting down
25-Feb-2014 17:23:42.999 client.c:2356: unexpected error:
25-Feb-2014 17:23:42.999 isc_socket_accept() failed: shutting down
25-Feb-2014 17:23:50.022 unable to signal parent that we otherwise started successfully.
25-Feb-2014 17:23:50.022 exiting (due to early fatal error)



Additional info:

There is no other error
Comment 1 Tomáš Hozza 2014-02-25 14:31:25 UTC 
Can you please attach /var/named/data/named.run ? thanks
Comment 2 emoziko 2014-02-26 07:25:29 UTC 
Created attachment 867818 [details]
named.run file
Comment 3 emoziko 2014-02-26 07:28:32 UTC 
Hello Tomas. I attached named.run file, its compressed in tar.gz because it's more than 40MB.

I think that this issue related to systemd, because bind starts with "named" command, but not with service or systemctl command.
Comment 4 Tomáš Hozza 2014-02-26 07:45:39 UTC 
(In reply to emoziko from comment #3)
> Hello Tomas. I attached named.run file, its compressed in tar.gz because
> it's more than 40MB.
> 
> I think that this issue related to systemd, because bind starts with "named"
> command, but not with service or systemctl command.

Hi.

Thank you for the BIND log. I just want to make sure and investigate the log
file thoroughly. I'll drop a comment here when I have some new information...
Comment 5 Tomáš Hozza 2014-02-26 15:36:20 UTC 
So the problem is that it takes too long for BIND to start up when loading
such a number of zones. After a default timeout systemd kills the process.

You should copy the default named.service file to a /etc/systemd/system dir
which has higher priority and specify custom timeout for the startup.

1. cp /usr/lib/systemd/system/named.service /etc/systemd/system/
2. edit the /etc/systemd/system/named.service and add 'TimeoutStartSec='
   option there or read 'man systemd.service' for more information.

I noticed that your server sends NOTIFY for each zone it loads. If you don't
have any slave servers configured I would advise you to use 'notify no;'
statement in the options section of named.conf to stop the server from sending
NOTIFY.

From my point of view this is NOT A BUG in named nor in systemd. I'll wait
for your confirmation that some of described options worked for you.
Comment 6 emoziko 2014-02-27 06:50:43 UTC 
Hello Tomas.

I did as you said and problem is solved. TimeoutStartSec= option helped me

Yes I think it's not a bug and sorry for that. Thank you for your  help and time.
Comment 7 Tomáš Hozza 2014-02-27 08:09:40 UTC 
(In reply to emoziko from comment #6)
> Hello Tomas.
> 
> I did as you said and problem is solved. TimeoutStartSec= option helped me
> 
> Yes I think it's not a bug and sorry for that. Thank you for your  help and
> time.

No problem. You're welcome. I'm glad that it helped....