Your Web Host Might Cause Your Website to Be Broken In Trying to Secure It

A lot of our business cleaning up hacked websites involves people hiring us to clean up their website after someone else had been hired to do that but failed to even attempt to do things properly, so we are not surprised at most of what we see done instead of doing that, but there is the occasional exception.

We were recently brought in to clean up a website after a web host was supposed to have handled that, but they didn’t seem to have made much attempt at that. What we found was that they had failed at doing some basic things that should have been easier parts of the cleanup. That included failing to remove the malicious JavaScript code that they knew a web based scanner had (correctly) identified was on the website, despite the being added to files on the website in a non-obfuscated way, so a simple search of the files would have found it. They also had failed to enable the archiving of logging of requests to the website, which would have been done through cPanel and would have made would have made it easier to spot the malicious file that was allowing a hacker to continue to access the website.

What stood out was something they claimed they had done:

Another thing I did to help prevent this from happening again in the future, I created a two-user system for your WordPress admin. What this does is it assigns a front-end user that has just enough permissions for the site to display, but not enough for people visiting the site to make changes, without logging in. It also has a back-end user that applies whenever someone logs in, that has all permissions.

Many of those that use WordPress that are not even familiar with security would probably find that odd sounding. You normally don’t have to be logged in to WordPress to view the website and when not logged in, not surprisingly, you can’t make changes to the website, normally. This website was a run of the mill WordPress website, so that was all true for it.

When we went to see what had actually been done we found that there was only one WordPress user, so either they were not referring to a WordPress user or they hadn’t done what they said. One possible explanation was that they were referring to another database user with only read access, which would not really match what they described and seem like a bad idea as WordPress isn’t designed for doing something like that.

In checking into this we found they were referring to a database user and they had created a bit of a mess.

While it looks like they created a database user that was only supposed to be read only as the two database user name were identical save for a “ro” added to the end of one, that wasn’t actually what they had created. The read only user has the following privileges DELETE, INSERT, SELECT, UPDATE. As you might be able to guess from some of those names, that user doesn’t just have the ability to read things, but make significant changes.

It turns out though they didn’t set up a dual user situation, so only that user was being used. That was a problem since privileges the user was missing were needed by WordPress in the course doing normal functioning. In the error logging we could see this had caused actions trying to be taken by a security plugin to not work.

That seems like a good reason to be wary of having a web host clean up a hacked website, but tomorrow we are going to be discussing another recent instance where a prominent security company failed to clean up a website the first time, then on the second go around managed to break the website and then break it some more.

Making an Unnecessary Change to a Website That Breaks Updates is Not Good for Security

There is a nearly endless amount of bad security advice for websites, so someone has to try hard to make theirs stands out, but that is what something we happened to run across recently from a company named ENDURTECH did.

Their post, https://endurtech.com/setting-proper-chmod-permissions-for-wordpress-wp-config-php-and-htaccess/, suggested that you should change the permissions on a couple of WordPress files to the “proper” permissions:

Set CHMOD Permissions to 444 on the following files:

  • .htaccess
  • wp-config.php

Those are not the proper permissions (if they were, you would assume that WordPress would set them that way for you) and they don’t make sense from a security perspective seeing as permissions only come in to play if someone has access to the files. In a normal hosting setup the only people that would have access to the files would also have permission to change the files permissions, so if you where to change those as suggested there, which would restrict editing the files, then those with access could change the permissions to be able to edit the files again, so this doesn’t provide a real benefit for most websites.

Bad advice is very common, what made this stand out is what is stated before that in the post:

Please note that doing as suggested within this article will no doubt eventually cause issues with WordPress plugin updates and maybe even WordPress core updates.

This is because these files are no longer “editable“.  Great for security, bad for updates.

Just keep this in mind and visit your website from time to time to make sure that your updates are completing correctly

Keeping software updated will actually have a positive impact on security, so they are suggesting doing something that isn’t useful that by their own admission makes something useful harder, which is bad idea.

GoDaddy Says That Version of PHP for Which Support Ended 3 Years Ago Meets Their Stability and Security Requirements

You would think that if a web host owned a security company they would be better than other web hosts when it comes to security. With GoDaddy that isn’t the case, though that might be explained by the fact that the security company they own Sucuri, seems to be completely incompetent. As yet another example of the security issues with GoDaddy, while dealing with a support issue on a website hosted with them we found that they were making this claim about PHP 5.4 on the Programming Languages page of their control panel on the website we were working on:

PHP version 5.4 is available and meets our stability and security requirements.

Support for PHP 5.4 ended in September of 2015.

To make thing more confusing if you click the question mark icon next to radio selector to use that version of PHP on the page a message box appears that states:

Version 5.4 is no longer actively supported.

So is the first claim inaccurate or do they have really low standards for “stability and security”?

Bad False Positives from Wordfence Security and Quttera Web Malware Scanner WordPress Plugins

We often have people contact us that believe that a claim that their website has been hacked is false because they ran a scanner over and it didn’t find anything. We are not really sure why they don’t ask for the evidence behind the claim and try to see if they can confirm if that is accurate or not instead of running a scanner over the website, but considering they are not doing that it might not be surprising that they are instead doing something that is likely to not produce great results.

One problem is that the even if the scanner is effective at what is attempting to scan for, it may not be able to detect the type of issue that lead to claim that the website is hacked. Let’s say a web host detects a malicious file on the website, well that probably would be be something that a scan of the website’s pages from the outside would never detect.

Another problem is lack of evidence that various scanners are actually effective at what they are attempting to scan for and from our own experience, plenty of evidence that they are not effective. One area where we have seen evidence of that going back many years is with really bad false positives that indicate that these scanners are incredibly crude, so crude in fact that if we weren’t well aware of how bad the security industry is, we would have a hard time believing that they were even occurring. Below are a couple of them in WordPress plugins that we recently ran across that show the current poor state of such tools.

Quttera Web Malware Scanner

The first comes from the plugin Quttera Web Malware Scanner, which has 10,000+ active install according to wordpress.org. In recent thread on the support forum for that someone mentioned getting a false positive for what is quite common code. The plugin will warn when matching “RewriteRule ^(.*)$ h” in a .htaccess file, which would match when do some fairly common rewriting of URLs. Just doing that rewriting is not in any way malicious. The developer’s explanation for that wasn’t that this was a mistake, but that:

We mark it as suspicious because there are multiple malware instances utilizing this technique to steal/redirect traffic from infected websites.

Simply because malware uses common coding isn’t a good reason to flag any usage of it and that will necessarily cause the results of a scanner to be of limited use.

Making it seem like the developer really doesn’t know what they are doing in general, the description for that detection is “Detected suspicious JavaScript redirection”, which makes no sense considering that type of code has nothing to do with JavaScript.

Wordfence Security

The second instance of this involves a much more popular plugin Wordfence Security, which has 2+ million active installs according to wordpress.org, that we have frequently seen people believe is much more capable than it really is (sometimes they ignored evidence right before their eyes to continue to believe that).

A thread on the support forum of the plugin Ultimate Member was recently started with:

Wordfence seems to think there is a malware URL somewhere in the file class-um-mobile-detect.php:

* File contains suspected malware URL: wp-content/plugins/ultimate-member/includes/lib/mobiledetect/class-um-mobile-detect.php

but on comparison, the file’s contents are exactly the same as the latest file offered on https://ultimatemember.com

Can someone comment?

In follow to a question by the developer of the mentioned plugin, the original poster wrote:

I’m using 2.0.23 but as I’ve said the file in question is identical to the one found in the latest version. So as I thought it is a false positive. Maybe Wordfence doubled up on UM after the latest malware exploit.

In reality it was just that Wordfence’s scanner incredibly crude as hinted at by another reply in the thread:

It is caused by the URL: “http://www.vonino.eu/tablets” which was reported to contain malware.

In my file, it’s only mentioned in a comment so I guess it’s safe.

What that is referring to is the following line in the file /wp-content/plugins/ultimate-member/includes/lib/mobiledetect/class-um-mobile-detect.php:

340
// Vonino Tablets - http://www.vonino.eu/tablets

Currently the domain vonino.eu is being flagged by Google as malicious:

That doesn’t in any way make a file that includes the domain in a commented out line in the code, which can’t run, in any way malicious. If the developer’s of Wordfence Security cared at all they could easily avoid that false positive, but considering they can get away with much worse it isn’t surprising they wouldn’t care about that. That also leaves more responsible plugin developers to have to deal with the fallout from those false claims.

The Developer of Cerber Security, Antispam & Malware Scan Gives Out Bad Advice To Push Their Plugin

When it comes the security industry around WordPress unfortunately there are many people that either don’t know what they are talking about or are intentionally peddling bad information to push products and services that provide little to no protection, while making things harder for companies that are actually doing the hard work to actually improve security.

We often run into examples of this even when we aren’t looking for them. We ran into another example just the other day when we went to look around for some information while working on a post about running into a problem with contact form due to WordPress’ REST API being disabled. That lead us to an example of someone at best not knowing what they are talking about when it comes to the basics of WordPress security while being the developer a security plugin, Cerber Security, Antispam & Malware Scan, that currently has 90,000+ active installs according to WordPress.org.

A big tell that developer doesn’t have a basic clue as to security surrounding WordPress is that a main feature of their plugin is blocking brute force attacks despite the fact that those are not happening. They also make this brute force related claim in the marketing materials for plugin:

By default, WordPress allows unlimited login attempts through the login form, XML-RPC or by sending special cookies. This allows passwords to be cracked with relative ease via brute force attack.

Saying that brute force attacks could crack a password relative ease is belied by the number of login attempts needed to actually test out all of the password combinations. Here is what we wrote about that previously:

To understand how you can tell that these brute force attacks are not happening, it helps to start by looking at what a brute force attack involves. A brute force attack does not refer to just any malicious login attempt, it involves trying to login by trying all possible passwords until the correct one is found, hence the “brute force” portion of the name. To give you an idea how many login attempts that would take, let’s use the example of a password made up of numbers and letters (upper case and lower case), but no special characters. Below are the number of possible passwords with passwords of various lengths:

  • 6 characters long: Over 56 billion possible combinations (or exactly 56,800,235,584)
  • 8 characters long: Over 218 trillion possible combinations (218,340,105,584,896)
  • 10 characters long: Over 839 quadrillion possible combinations  (839,299,365,868,340,224)
  • 12 characters long: Over 3 sextillion possible combinations  (3,226,266,762,397,899,821,056)

The post that we had run across was “Why it’s important to restrict access to the WP REST API”. The post is riddled with errors, for example citing someone as having discovered a vulnerability they didn’t.

The general problem was that they were suggesting disabling the REST API, which not at all coincidentally they touted their plugin did, because there could be security issues with it since it is new. But that is true of anything. In reality the vulnerability they discussed in the post actually showed how WordPress does a good job in handling security in one important way, since the auto update mechanism that has been in WordPress 3.7 allows the vast majority of WordPress website to be updated to a new security release in a very short time. Normally WordPress checks for updates every 12 hours and that can be shortened when a security update is being released, so most of the websites would likely have been updated in around 12 hours. With this vulnerability there was no evidence of it being exploited until after it was disclosed that it had been fixed a week after the version that fixed it was released (while the information on this vulnerability was held back for a week, other security updates were mentioned when it was released).

The developer though put forward a very different impression:

Unfortunately, the REST API bug had not yet been fixed. That leaves unprotected millions of websites around the world. It’s hard to believe but updating WordPress on shared hostings may take up to several weeks. How many websites have been hacked and infected?

That it may take several weeks to for WordPress on shared hosting to update is actually hard to believe, since it doesn’t appear to be true and no evidence was presented to back up a claim even they claim is counter-intuitive. The developer provides no evidence that any websites were hacked before the vulnerability was disclosed as having been fixed a week before, which as far as we are aware they couldn’t have since it doesn’t appear any were. That all probably shouldn’t be surprising since the developer apparently had never checked to see if brute force attacks were actually happening before building a plugin to protect against that.

For website where the auto update mechanism was disabled or didn’t work they did get mildy hacked due to this vulnerability, but that is the only vulnerability in more than a decade that we are aware of where there was any sizable number of websites hacked (in that time outdated WordPress installation have been frequently falsely blamed for the hacking of websites by security companies that either didn’t know what they were talking about or intentionally lying to get themselves press coverage). So disabling the REST API subsequent to this vulnerability being fixed has not actually improve the security of websites in any meaningful way.

There also was the issue of the developer conflating bugs and security vulnerabilities, which is important since having a lot of bugs fixed in something doesn’t mean that there was security risk.

The downside of disabling the REST API can be seen in that, like with the other plugin we mentioned in the post from earlier this week, this plugin can cause Contact Form 7 based forms to stop functioning. This exactly the kind of downside that often isn’t considered when people indiscriminately use WordPress security plugins and services without finding out first if there is any evidence that they provide effective protection. In this case what makes this stand out more to us is that our Plugin Vulnerabilities plugin, which is designed to help protect against a real issue, is much less popular than this plugin. It could be worse though, as another security plugin just designed to protect against brute force attacks has 2+ million active installs according to wordpress.org and it not only doesn’t protect against a real threat, but contains a security vulnerability of its own.

Disabling WordPress’ REST API Can Cause Contact Form 7 to Not Work

In our work for our Plugin Vulnerabilities service we frequently need to contact developers of WordPress plugins to let them know about security vulnerabilities in their plugins (either that we have discovered or that others have disclosed) and that often means submitting messages through contact forms (not surprisingly these are often handled by WordPress plugins). We have all too frequently run into situations where the contact forms didn’t work, which seems like a good reason for people managing websites with a low volume of contacts to periodically make sure that contact forms work, otherwise you could be missing out on messages.

In a recent instance of this, a loading graphic showed up after hitting Send and then that didn’t change to a message about the form being successfully sent. Pulling up the web browser’s console showed an error:

Failed to load resource: the server responded with a status of 401 (Unauthorized)

The page that related to was /wp-json/contact-form-7/v1/contact-forms/193/feedback, which would indicate the Contact Form 7 plugin was being used to handle the contact form. Visiting that page showed the following message:

{“code”:”rest_cannot_access”,”message”:”DRA: Only authenticated users can access the REST API.”,”data”:{“status”:401}}

Based on that it seems that disabling of WordPress’ REST API for those not logged in to WordPress caused the contact form to not work. A quick search showed that message was generated by another plugin, Disable REST API, which as the name suggest disables WordPress’ REST API.

As this shows, using something like that that disables the REST API can have some serious downsides. Not surprisingly for us, while looking into this we found someone in the WordPress security industry that doesn’t seem to have a clue about WordPress security pushing disabling it (and promoting using their plugin to do it), which we will discuss in a follow up post.

Excessive Debug Log Files Can Slow Down Zen Cart Admin Area

We recently ran into an issue while working on an upgrade of Zen Cart website, which seems worth sharing in case someone else runs into a similar issue.

We had first done a test of the upgrade using a copy of the website we made through FTP access and placed into a directory on the existing website. Everything worked fine with that test copy of the upgrade. Then after the production website was upgraded the client was noticing that there very long load times for pages in the admin area of Zen Cart. That wasn’t happening on the test copy and apparently that hadn’t been happening on the production website before the upgrade (though we couldn’t confirm that). Seeing as the two websites should be identical that didn’t seem to make sense.

We then stumbled in to the answer. In trying to debug things we found that over FTP we couldn’t see all of the debug log files in the /logs/ directory for the website, as there was a limit of 8,000 items being shown, which also meant the test copy only started with that many items. We couldn’t see how many files there were in the production website’s directory as in the file manager in the cPanel control panel for the website the amount of files was apparently too many for it to be able to display any files in the directory. When we temporarily replaced the /logs/ directory on the production website to be able to see the latest entries, we found that the page load time in the admin area were no longer so slow.

What looks to have been happening is that when visiting an admin page a new file was being created to warn that the setting for where to place log files of the parse time was invalid and the amount of files in the directory was slowing creating that file, leading to the slow load times. So clearing out the old debug log files would solve such a situation, but if there are lots of debug files being generated dealing with the cause of those is important as the files count will just start growing again if that isn’t dealt with.

BitNinja Makes Up Zero-Day Attack

The terribleness of security companies never ends. The latest example of that is something we ran across today while looking in to a claim that outdated software was the cause of a security issue on a server. What had been pointed to as evidence of that was a report from a security company named BitNinja. That report was claiming that there was malicious activity based on emails being sent from software on a website, but based on the information provided there was nothing that we could see that would indicate if there really was an issue or if there was a false positive happening (it would seem that the company doesn’t have a good understanding of what information is important to determine that sort of thing).

In looking over BitNinja we quickly ran across evidence of them spreading false information, which happened to involve a topic we just discussed earlier today, exploitation of a recently fixed vulnerability in MODX. The title of a blog post on their website made a striking claim about that, “Critical zero-day vulnerability in MODX Revolution patched by BitNinja WAF”. A zero-day vulnerability refers to a vulnerability that is being exploited before the developer is aware of it, so they have had zero-days to fix it. That obviously is quite concerning since doing the security basic of keeping software up to date wouldn’t protect against and if there was a security system that could protect against such a situation it would be useful.

With a website that had been hacked through that vulnerability the attempts to exploit it on that website started about a week after the vulnerability was fixed, with the first attempts logged on July 19. There was nothing we saw in looking into the situation that would indicate that that this was a zero-day vulnerability.

BitNinja seems to either not have any idea what they are talking about or intentionally misleading people as their claim that this is zero-day vulnerability is based on spotting exploitation attempts two weeks after a fix for the vulnerability had been released:

At 26th July at 6 PM, the flow has been started according to our data. This botnet is really aggressive, as, in the first 6 hours, we detected almost 13.000 attacks!

They also were quite behind in even spotting the attacks, which doesn’t say great things about them either.

Blaming the Victim

Looking at their About Us page a couple of things stood out to us, one of them being them starting with a claim of near equivalency between hackers and people running web servers:

We believe every server owner is responsible for their servers. If they have been hacked – and used for cybercrime – the owner is almost as guilty as the hacker is.

There also is the basis of their business that doesn’t seem to be from a security background, but one of a web host not being able to maintain their servers:

We couldn’t ensure the security of our servers beyond applying continuous updates. To make matters worse, we started losing customers after a series of downtimes. We quickly realized that server security is not a question of a single component but is about several components working together to harden a server. This inspired us to create BitNinja, an all-in-one security solution designed for hosting providers.

They don’t make any claim to having security expertise on that page (not that it would mean much based on what we have seen of security companies making such claims).

Vulnerability in Older Versions of MODX Being Exploited

Quite often with hacked websites outdated software is pointed to as the source of the hack. That is usually a claim that is made without any knowledge if the claim is actually true. Many security companies that market themselves as having unique expertise in dealing with hacked websites don’t even attempt to determine how websites are hacked, despite that being one of the three key components of a proper cleanup, so they would have no idea what the cause might be. Often times these companies don’t seem to even have a cursory knowledge of what they are talking about either, as an example, one well known security company, Sucuri, once told people to update software despite it being well known that the vulnerability being exploited in the software was in the then current version of the software (that kind of thing somehow never stopped journalists from repeating misleading and false claims made by that company or people claiming that they are a reputable company).

From what we have seen those baseless claims are usually easy to spot as there usually isn’t even a specific vulnerability that is pointed to as being the cause of the hack, which should be something known if someone has actually done the work to determine the source of the hack and determined it was outdated software.

As example of finding out that outdated software was actually the cause of a hack, we were recently brought in to clean up a hacked MODX website. MODX websites have not been a common type of website needing cleanups from us recently, so the software in use on the website was of some note right away.

In trying to determine how a website was hacked the logging is probably the most important resource, but the files can often tell you a lot, and both of them can work together to speed up the process. In the case of this website there was an obviously malicious file named dbs.php in the root directory of the website. That file had also had a number of POST requests made to it, which are requests that contain additional data and of which most requests sent by hackers are of that type, sent to it shortly before we started the cleanup. Looking back at the logging to where that file was first requested we found it in a set of requests sent by an IP address from Ukraine:

134.249.50.5 – – [19/Jul/2018:19:55:23 -0400] “GET / HTTP/1.1” 403 134 “-” “Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36”
134.249.50.5 – – [19/Jul/2018:19:55:23 -0400] “POST /connectors/system/phpthumb.php HTTP/1.1” 403 134 “-” “Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36”
134.249.50.5 – – [19/Jul/2018:19:55:24 -0400] “GET /dbs.php HTTP/1.1” 403 134 “-” “Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36”

The first request there is for the homepage of the website. The second one sends a POST request to a file /connectors/system/phpthumb.php. Finally there is a request for the dbs.php file. Based on that, it would appear that the file phpthumb.php would be the vector for adding the dbs.php file.

In reviewing the file phpthumb.php there wasn’t anything in the file itself that looked like a vulnerability that would permit uploading a file as that series of requests would indicate was what the hacker would be attempting to do. In fact the file only contained four lines of code that just called on code in other files:

define('MODX_CONNECTOR_INCLUDED', 1);
require_once dirname(dirname(__FILE__)).'/index.php';
$_SERVER['HTTP_MODAUTH'] = $modx->user->getUserToken($modx->context->get('key'));
$modx->request->handleRequest(array('location' => 'system','action' => 'phpthumb'));

Instead of digging through more code at that point we instead did a web search for “/connectors/system/phpthumb.php” and though that we got pointed to the issue. There was a post of the details of a vulnerability that matched what we had seen that was published on July 13 and what seems more important, code for exploiting the vulnerability that was released on July 18. On this website the first attempt to exploit it was one July 19, so it would seem the code to exploit it was quickly utilized by hackers.

That vulnerability had been fixed in version 2.6.5 of MODX, which was released on July 11, and the developers provided clear notice of the need to update due to security fixed in it. Writing in the release announcement

Today we released MODX Revolution 2.6.5. It contains fixes for two critical security vulnerabilities affecting all versions at or prior to 2.6.4. Upgrading to 2.6.5 should be considered mandatory.

and

Upgrading is Critical

Revolution 2.6.5 contains critical security enhancements, you should upgrade to 2.6.5 now. See below for more info.

We cannot stress the importance of diligently upgrading to the latest version of MODX enough. While no software is 100% secure, powering your site with the most current version usually helps protect you from hackers that rely on exploiting outdated software. If you’re not sure what version of MODX Revolution you’re running, log into your website Manager. If the version number doesn’t appear in the top left-hand corner of the Manager, go to Manage>Reports>System Info.

The two vulnerabilities refer to the ability to upload files and to remove files/directories. From the post with the details of the vulnerability it sounds like in version 2.5.1 to 2.6.4 the ability to exploit the file upload vulnerability would be more restricted than was the case with the website were dealing, which was running 2.4.1.

Cleaning Up After This Hack

On this website the hacker had done quite a number on it. The .htaccess file in the root directory had been removed, leading to all the pages other than the homepage to no longer being functional. That seems to have been done to remove any restriction in the .htaccess file that would have blocked the hacker from sending requests to the malicious files they were uploading.  When trying to go to the admin area you would be redirected to another website due to the content of all of the JavaScript files on the website being replaced with malicious code.

The best option to clean up after this would be restore a clean back up from before the hack (making sure that all of the existing files are removed during the restoration). Seeing as the vulnerability wasn’t disclosed until July 13, a backup from before then would be a good option. You might be able to get way with one from before July 18 as well. A review of the logging by someone familiar with all of this would likely be able to confirm when the hacker hacked the website.

From what we could see from that website, it would appear that there are likely multiple hackers exploiting this vulnerability and doing different things, so it wouldn’t be possible to provide general instruction on what to remove from the website to clean up if there isn’t a backup available (though based on past experience that won’t necessarily stop someone from claiming to provide that and unintentionally or intentionally leading people astray).

If you are looking for a professional cleanup from this or any other hack of a MODX website we provide a service for that. We can also upgrade MODX for you.

Fixing Zen Cart Admin Login Redirect Loop Caused By Forcing Website to Be Served Over HTTPS

We recently had someone come to us for Zen Cart support after they could no longer log in to the admin area of Zen Cart after their web host configured it so their website would always be served over HTTPS. When they tried to log in they were redirected back to the log in page without any error message being shown. While there are some other issues that can cause that same type of redirect to occur, in the situation where that change to serve the website over HTTPS has been made, what we found would fix this is to update the configuration file for the admin area, which is located /includes/configure.php in side of the admin directory (whose name varies from website to website), to use the HTTPS address for the website.

The relevant portion of the configuration file for a website using a recent version of Zen Cart is below:

/**
 * Enter the domain for your Admin URL. If you have SSL, enter the correct https address in the HTTP_SERVER setting, instead of just an http address.
 */
define('HTTP_SERVER', 'http://www.example.com);
/**
 * Note about HTTPS_SERVER:
 * There is no longer an HTTPS_SERVER setting for the Admin. Instead, put your SSL URL in the HTTP_SERVER setting above.
 */

Changing the “HTTP_SERVER” setting to start https:// instead of http:// resolves this as the proper address is used when handling the log in.