Segmentation Fault - cron.php

[logart]

Hello,

Linux Centos 5.4
Sugar: v5.5.2 CE
php 5.2.10-1.el5.centos
Apache: 2.2.3
MySQL: 5.0.77

I am having an issue with the schedulers. My cron is failing with a "Segmentation fault", and I think it is happening during pollMonitoredInboxesForBouncedCampaignEmails. After the fault, the Bounced Email scheduler is stuck in "In Progress". Any ideas?

steve

Here is the log:

# /usr/bin/php -c /etc/php.ini -f cron.php;tail -f sugarcrm.log 
Segmentation fault
Wed May 12 18:11:05 2010 [30453][1][INFO] Query:SELECT schedulers_times.* FROM schedulers_times  WHERE schedulers_times.id = 'd0fe36e3-4625-ac2a-2ec6-4beab6d7df58'  AND schedulers_times.deleted=0 LIMIT 0,1
Wed May 12 18:11:05 2010 [30453][1][INFO] Query Execution Time:0.0025088787078857
Wed May 12 18:11:05 2010 [30453][1][DEBUG] ----->SchedulersJob firing pollMonitoredInboxesForBouncedCampaignEmails
Wed May 12 18:11:05 2010 [30453][1][INFO] ----->Scheduler job of type pollMonitoredInboxesForBouncedCampaignEmails()
Wed May 12 18:11:05 2010 [30453][1][INFO] Query:SELECT id FROM inbound_email WHERE deleted=0 AND status='Active' AND mailbox_type='bounce'
Wed May 12 18:11:05 2010 [30453][1][INFO] Query Execution Time:0.0049028396606445
Wed May 12 18:11:05 2010 [30453][1][DEBUG] Retrieve InboundEmail : SELECT inbound_email.* FROM inbound_email  WHERE inbound_email.id = '4bceaa85-5239-04bd-19bf-4be0b1c99367'  AND inbound_email.deleted=0
Wed May 12 18:11:05 2010 [30453][1][DEBUG] Limit Query:SELECT inbound_email.* FROM inbound_email  WHERE inbound_email.id = '4bceaa85-5239-04bd-19bf-4be0b1c99367'  AND inbound_email.deleted=0 Start: 0 count: 1
Wed May 12 18:11:05 2010 [30453][1][INFO] Query:SELECT inbound_email.* FROM inbound_email  WHERE inbound_email.id = '4bceaa85-5239-04bd-19bf-4be0b1c99367'  AND inbound_email.deleted=0 LIMIT 0,1
Wed May 12 18:11:05 2010 [30453][1][INFO] Query Execution Time:0.0041821002960205

[logart]

Just a side note to this:

This seg fault is preventing the rest of the schedulers from running. My mail queue will not function through cron with the bounce scheduler failing.

[andopes]

Take a look into error_log of apache. Hopefully there is any error there related to such issue.

Cheers

[logart]

I don't see any error messages related to that issue. The php executable is seg-faulting, so it probably dies before it opens the log.

I have LOTS of other error messages in my logs though.

[richmarotta1]

Did anyone ever figure out how to fix this?

My configuration is :
PHP 5.2.10 (no other ph version installed)
mysql 5.0.84
Sugar CE 6.1.3
Centos 5

The bouncedemail job hangs because of this.

I noticed that I don't get the segmentation fault if there is a "In Progress" record in schedule_times. However it won't schedule a new job if the "In Progress" record is there. If you run it with out the "In Progress" record already existing, it creates the "In Progress" record and then you get the segmentation error .

Thx

[andopes]

Make sure your php is configured to log errors.
Additionally take a look at sugarcrm.log to identify if there is any useful error message.

Regards

[richmarotta1]

bom dia,

Sugar is set to debug mode and PHP is set to E_ALL.



Nothing in PHP log.

In the sugar log, when it doesn't seg. fault ,the sugar log stops with the line below. Remember it only seg. faults when there is not a "In Progress" record in the table schedule_time. If it find a "In Progress" record that has not met the time limit it will do nothing and continue on its merry way.

Here is the last line from the sugar log when it doesn't seg. fault:
Wed Apr 27 10:25:35 2011 [26308][1][INFO] Bounced campaign scheduler connected to mail server id: 14788f06-ac83-8b6d-fcae-4da4e0732e05

I have spent a lot of time and done a lot of research and testing this bug. It is very irritating as it is the last piece of my implementation.

Here is additional info.:

I have attached the output of the command "strace -ecf cron.php". You'll see the seg. fault at the end.
I have also attached a copy of the sugarcrm.log that is created when I run that command and it eg. faults.

In reading other info on the web there was a lot of mentions of a bug in the mysqlcli on debian OS. I believe it was resolved with a patch.

I am on Centos 5 and I have the latest supported version of mysql which is 5.084.
I updated and upgraded all my OS components using Yum.

Any help is appreciated. Do you now if the other person who started this thread ever resolve the issue?


Obrigado,
Rich

Falo um pouco de português. minha esposa é um paulista. seus pais têm uma casa em Cabo Frio. Nós amamos o Brasil.

[noot]

Hi.

Did you finally solved the issue? I have the same issue on CentOS 5.7 with SugarCRM CE 6.2.3

I'm missing the strace output. I would like to have a look at it. Please post it one more time.

Regards,
Arnaud

[noot]

Some additional findings:

The cron job didn't show any error in the apache error log, but when I switched to run the job using wget the server log started to show a segmentation fault. Then I was able to match one line from debug sugarcrm.log with the same time stamp of 17:26:04:

Mon Oct 10 17:26:02 2011 [3656][1][DEBUG] Hook called: InboundEmail::after_retrieve
Mon Oct 10 17:26:04 2011 [3656][1][INFO] Bounced campaign scheduler connected to mail server id: 9ae2ac87-2847-5645-ed0f-4e8e62453a23
Mon Oct 10 17:26:06 2011 [5999][-none-][DEBUG] current_language is: en_us

Can we safely say now that connection to Google's mail server causes the seg-fault?