Jump to content


kiwiphnx

Member Since 21 Nov 2017
Offline Last Active May 15 2018 07:01 AM
-----

Posts I've Made

In Topic: So Johnny died...

13 May 2018 - 07:27 AM

Would it help to run Johnny (or, more appropriately, "Kenny") off a volume or filesystem that supports snapshotting, e.g. OverlayFS or LVM; the idea being to snapshot the OS daily and hosting accounts hourly, allowing most files to be kelp read-only and rolling back the snapshot upon server crash?

In Topic: [Solved] Suspended: kiwiphnx

12 May 2018 - 10:40 AM

We scanned the site for malware and got it up again on the new host. Long story short, WordPress or one of its plugins may have broken itself on the last update.

Although I haven't yet connected the database on the new host; loading the site from cache revealed that the WordFence firewall is generating a request of around 250 bytes with every mouse movement. Perhaps this is a contributing factor to the problem?

In Topic: [Solved] Suspended: kiwiphnx

11 May 2018 - 10:35 AM

Thanks again. The medium-term goal for us is to migrate the site to a better-suited CMS such as Concrete5 or Drupal; WordPress was never really intended to be a permanent solution. At this point, however, we have committed ourselves to migrating the website to paid hosting on a shared New Zealand-hosted VPS; so whatever we do now will only be a very short term solution (two weeks tops). I have already updated CloudFlare DNS to point the www and static subdomains to the new server, so that will only leave small stuff like the URL shortener and some mail forwarding rules.

I apologise again for the load on your servers, whatever the root cause turns out to be; and Phoenix Adviser Group Ltd. thanks you for providing the service that you do.
During Phoenix's time on Tommy, I have tried everything I can think of to tune performance (believe me, 10s+ page loading times are no fun for us either); including deploying CloudFlare CDN, agressively caching the HTML output from WordPress, and writing a very complex .htaccess ruleset to try to bypass WordPress for as many requests as possible.

I have even suggested ways in which HeloiHost can increase the reliability and uptime of all your servers by, e.g. deploying a high-performance reverse proxy (e.g. Varnish) in front of them; potentially allowing Apache restarts to be overlapped (on different ports) on the backend. I think what HelioHost does is very noble, and I want you to be as successful as possible.

In Topic: [Solved] Suspended: kiwiphnx

11 May 2018 - 04:53 AM

I notice the account has become suspended again; so I can only guess something has happened quite recently to cause the unacceptably-high load (WordPress exploit or misbehaving plugin, perhaps?).  Fortunately, I did have sufficient time to create and pull the backup, so the matter is now much less urgent.

 

With your help, I would suggest we reset both the document root and .htaccess; and rebuild a static copy of the site using the contents of /www/static/uploads and /www/static/cache/cache-enabler/; as it will take me at least a few days to rebuild the site on the new host and we cannot afford to lose our web presence during that time.

 

If you do find any insights into what has caused the unusual load, it would be greatly appreciated if you share them with me; as my strategy at present is to assume the site has been compromised and to rebuild it from scratch.


In Topic: [Solved] Suspended: kiwiphnx

11 May 2018 - 02:03 AM

I've retrieved and deleted the backup.  Now if we can only figure out what is causing the account to generate such a high load.  Were we the target of an attack; Is the WordPress caching not working properly; or is there perhaps an error in the .htaccess rules?

 

I want to work with you to serve at least a static version of the website while I work to get the site live on our new host; the biggest problem I have right now (and, perhaps, this is the problem) is that the mod_rewrite ruleset in .htaccess has grown so complex (primarily as a result of having to host more than one web on the account) that I am not entirely sure I have a complete understanding of them anymore.