8 posts / 0 new
Last post
JAS's picture
JAS
Last seen: 6 months 3 weeks ago
Joined: 12/16/12
Posts: 256
Does anyone know why this website keeps going down?

For the last couple weeks, the C36IA website has gone down almost as much time as it has been up and running.  Just curious if anybody knows why this keeps happening?

Joel Schwartz
s/v Get-A-Grip
1993 Catalina 36 Hull #1259
Newport Beach, California

--
Joel Schwartz
s/v Get-A-Grip
1993 Catalina 36 Hull #1259
Newport Beach, California
Channel Islander's picture
Channel Islander
Last seen: 45 min 27 sec ago
Joined: 10/8/11
Posts: 378

Hi, 

In fact, the problems started six days ago when the hosting provider made changes to the platform. Since then the Apache webserver has been randomly shutting down. I've increased log level to trace3 but nothing shows before the exit via SIGTERM. It's very odd. I've been of course trying to reach out to the hosting provider but it's difficult to get a solid block of time to raise them, and especially when there's nothing to show. Still they may see something correlating. Meanwhile I guess I will just add a cron job to check every minute, or maybe there's a canned way to run the server as a daemon. Stay tuned.

--
Nick Tonkin
*Former* Website Administrator, C36/375IA
*Former* owner, C36 tr/fk #255, Santa Barbara, CA
ludo's picture
ludo
Last seen: 1 week 6 days ago
Joined: 3/14/16
Posts: 82

Do you monitor the system? Maybe some bottleneck issue on the memory or IO processing with the new system. I saw that issue with apache and the backend system is not performing correctly or when a DDoS is happening, it could be also one issue to monitor.

Do you use something like https://uptimerobot.com/ for downtime notification? Pretty useful free tool.

--
Ludovic François
​Hotel Catalina - Catalina 36 Hull #883
Marina Del Rey, CA
Haro's picture
Haro
Last seen: 2 hours 40 min ago
Joined: 11/7/14
Posts: 248

The modules on Drupal sometimes interfere and cause it to go off line.
Not all webhosting services can support Drupal.
 I use http://1and1.com for all of my Drupal hosting packages.
I have no affiliation with them.

Channel Islander's picture
Channel Islander
Last seen: 45 min 27 sec ago
Joined: 10/8/11
Posts: 378

Hi Haro,

Thanks for the insight. Do you have any specifics about what Drupal modules create problems?

As usual, this problem is vexing. It began last Thursday night, which is unrelated to any changes. Nothing was changed in the Drupal stack, the server was not rebooted (I mistakenly thought it has been) ... We just started getting the Apache server shutting down due to the "out-of-memory killer" that is installed on the host. FYI we are hosting the site ourselves now on Linode. Currently I am trying to install detailed monitoring of Apache so we can try to find out what's going on. I will also install Uptime Robot today (thanks Ludo, I didn't know that was available without cost), as well as write a cron script to check and retsart the server if down.

--
Nick Tonkin
*Former* Website Administrator, C36/375IA
*Former* owner, C36 tr/fk #255, Santa Barbara, CA
Haro's picture
Haro
Last seen: 2 hours 40 min ago
Joined: 11/7/14
Posts: 248

Nick,
I host my own Drupal on a RAID server that I set-up.
Having your own server is best way to have access to all of its settings.
There are changes that need to be made to its php.ini file and settings.php file to make them work smoothly.

I google the out of memeory Apache problem and found these links ( they called it OOM)
https://monicalent.com/blog/2013/03/27/avoiding-the-oom-killer-by-limiting-the-number-of-apache2-processes/
http://www.oracle.com/technetwork/articles/servers-storage-dev/oom-killer-1911807.html​

 

Channel Islander's picture
Channel Islander
Last seen: 45 min 27 sec ago
Joined: 10/8/11
Posts: 378

Morning Haro,

That was a helpful pointer. The default Apache config had 40 spare workers which is way more than we need. I lowered it to a smaller number, but noticed (as pointed out in the doc) that the memory consumption at rest didn't change -- Apache parcels it pout among workers apparently. Anyway the server ran OK until about 8pm PST when it was shut down again. Unfortunately once the server is down the server stats are not avaiilable through Longview so I was not able to go back and see if there was a spike. In the past there has been no evidence of weird requests though. Anyway it was disappointing to awake to another failure after the server adjustments. Today I really will just make a sledgehammer cron script to restart the server if it's down.

--
Nick Tonkin
*Former* Website Administrator, C36/375IA
*Former* owner, C36 tr/fk #255, Santa Barbara, CA
motarded421
Last seen: 9 months 3 days ago
Joined: 4/20/18
Posts: 39

The Drupalgheddon patch last week has caused me a lot of headache on sites at work. While I've never been a big fan of Drupal, these recent issues have moved me firmly in to the "I hate drupal and will never build another site based on it" ..
 

--
'86 Catalina 36
Work in progress!
Log in or register to post comments