I have witnessed at least two cases where site verification failed because provision could not restart apache.

Oddly enough, retrying the task fixed the issue.

I will try to have a more reliable way to reproduce.

CommentFileSizeAuthor
#8 aegir_verbose_restart_441864.patch1014 bytesanarcat
#7 aegir_verbose_restart_441864.patch1013 bytesanarcat

Comments

anarcat’s picture

anarcat commented
Status: Active » Fixed

So this is weird: i can't reproduce simply by creating and verifying a site. Last time I was able to reproduce, it was by deleting and reimporting a complete platform. I can't reproduce by doing that so I guess it was something in my environment.

For the record, even if this bug actually exists, there is a workaround: just retry the task. If this bug is reopened, we will need clear documentation on how to reproduce.

z.stolar’s picture

z.stolar commented
Status: Fixed » Active

I don't know how to reproduce, but I keep running into the error, even when retrying the task.
Here is the task's log:

Task starts processing
Running: php /Applications/drush/drush.php --root='/var/aegir/drupal-6.11' 'provision' 'verify' --backend
Drush bootstrap phase : _drush_bootstrap_drush()
Drush bootstrap phase : _drush_bootstrap_drupal_root()
Initialized Drupal 6.11 root directory at /var/aegir/drupal-6.11
Found command: provision verify
Initializing drush commandfile: provision_apache
Initializing drush commandfile: provision_drupal
Initializing drush commandfile: provision_mysql
Including /Users/XXX/.drush/provision/web_server/verify.provision.inc
Including /Users/XXX/.drush/provision/platform/verify.provision.inc
Including /Users/XXX/.drush/provision/dns_server/verify.provision.inc
Including /Users/XXX/.drush/provision/db_server/verify.provision.inc
Virtual host configuration path exists.
Virtual host configuration ownership of path has been changed to XXX.
Virtual host configuration permissions of path have been changed to 700.
Virtual host configuration path is writable.
Generating apache host configuration file platform_8.conf.
Web server could not be restarted. Changes might not be available until this has been done.
An error occurred at function : provision_apache_provision_verify
Removing task from hosting queue
An error occurred at function : hosting_hosting_task
Changes for hosting_hosting_task module have been rolled back.
The task is being retried and has been added to the hosting queue again
adrian’s picture

adrian commented

run apachectl configtest

anarcat’s picture

anarcat commented
Category: bug » support
Status: Active » Needs review

Please try the attached patch, which should show you the output of the command in the task log.

Please post the output of apachectl configtest.

Since I cannot reproduce the issue, I'm filing this as a support request for now.

adrian’s picture

adrian commented
Category: support » bug
Status: Needs review » Active

run apachectl configtest?

z.stolar’s picture

z.stolar commented

apachectl configtest: Syntax OK
@anarcat: where is the patch you were talking about?

Before you go any further in trying to assist here, I want to test a different apache config, as proposed by vertice on irc.

I'll update here.

anarcat’s picture

anarcat commented
StatusFileSize
new aegir_verbose_restart_441864.patch1013 bytes

Damn, forgot to add the patch.

anarcat’s picture

anarcat commented
StatusFileSize
new aegir_verbose_restart_441864.patch1014 bytes

The last patch introduces a parse error, here's a reroll.

Keep in mind the patch is only for debugging purposes: running the task as verbose or --debug would probably work also.

z.stolar’s picture

z.stolar commented
Category: bug » support
Status: Active » Fixed

OK. It appears my problem was lying between me and the keyboard, not beyond it...
I just didn't notice the necessary addition to the sudoers file:

    sudo visudo
    aegir ALL=NOPASSWD: /usr/sbin/apachectl

So the web server was waiting for a password each time, and couldn't get pass this step.

(thanks Adrian and anarcat for irc help)

adrian’s picture

adrian commented

run apachectl configtest

z.stolar’s picture

z.stolar commented

Why is it a bug? If I read the inline help properly, I wouldn't have ran into this issue.
apachectl configtest outputs Syntax OK.

anarcat’s picture

anarcat commented

Well, I'm not sure it's a bug either. Adrian seemed to think so, but I suspect he missed the last exchanges. This happened to me just after the upgrade from 0.1, but I have never seen this happened again, so I'll close this issue and put it back as a support request.

Status: Fixed » Closed (fixed)

Automatically closed -- issue fixed for 2 weeks with no activity.

tusharkanta’s picture

tusharkanta commented
Version: 5.x-0.2-alpha1 » 6.x-0.4-alpha3

I installed aegir following the install.txt (line by line) in centos. While verifying I got the following issues:

Undefined index: base_url
Undefined index: db_url
Web server could not be restarted. Changes might not be available until this has been done. (I have done all the necessary steps of adding aegir ALL=NOPASSWD: /usr/sbin/apachectl in /etc/sudoers)

Please go through the url to see the verification log status:

http://peppervillage.net/node/6

(username: peppervillageAdmin
password: peppervillageAdmin)

Please help me solve this issue at the earliest.

Thanks in advance.

tusharkanta’s picture

tusharkanta commented
Status: Closed (fixed) » Needs work
Anonymous’s picture

Anonymous (not verified) commented
Status: Needs work » Closed (fixed)

Please don't open old, closed tickets, and don't cross-post from other old, unrelated tickets.

Please open a new support request if needed. As a hint, did you read the CentOS hints and check these user-contributed notes regarding sudoers on CentOS.