Outages or not?

[DChan]

I have a monitoring service and I have been receiving notifications that there have been connection timed out and connection refused errors.

I checked the status code report in Analog Stats in cPanel for the .com(SSL) version of my website, it shows that there is thirty - 500 Internal server error and twenty five - 503 service temporarily unavailable for June.

Also, over the last of April and part of May, even though my traffic was up, my sales have become almost non-existent (not normal) and the bounce rate is high, some at 80% to 100% at times which I thought indicated there is a problem somewhere.

At the top of the Analog Stats page in cPanel in the "General Summary" section it shows that there have been "Failed requests: 2,222,152" for the month of June. Even my Google Search Console shows a spiking in the "Time spent downloading a page (in milliseconds)" going from a low of 41 to 157 and climbing.

I have been discussing this with my webhost. Even though I sent them a screen shot of the outages from the monitoring service, the status code Analog stats in cPanel, and the GSC graph, they say the following,

"There are 0 500 errors shown in the access logs."

"I'm not sure where its getting these from, I have to think its misreporting it, as mentioned a few times, there are NO 50* errors mention in any of the access logs, I've checked them all"

Does anyone have any ideas about what could be going on? And what would you do about this? I need to do something because I have lost most of my business within the last month, yet my traffic is up a little from normal.

Thanks in advance for your help.

[tangor]

As the web moves from http to https there will be some reporting errors ... however, what you have described does seem a bit extreme. I have not experienced this ... but thanks for the heads up! I will be watching in future to see if anything like describe starts up!

[DChan]

Thank you. my website was migrated from http to https a couple of years ago so that shouldn't have anything to do with it.

[phranque]

Have you examined the server access logs yourself?

[Dimitri]

Do you have access to the logs? It sounds internal server error. This can be high load average, impossibility to answer all requests, it can be database access problem, etc etc

[DChan]

I tried but I don't understand them. I am in the Raw Access of cPanel, downloaded the Jun 2019 and opened in Word...how would the error be listed?

What I really don't understand is why they are saying the logs don't show a 500 internal error but the Analog Stats do show the errors....?

[DChan]

I keep getting the same reply from the web host, "I checked all available logs, there are no reported 50x errors" so I don't know what to do from this point.

[DChan]

I don't know if this has anything to do with it or not but I did notice that when I click on "server information" in the right hand column of cPanel, and scrolling down the information of the window that comes up in the "Disk / (/)" column it has "97 %" in details and a red error button in the status column instead of a green button check mark.Is that an issue?

[DChan]

There has been a couple of times over the two years I have been with this web host that I have attempted to ftp updates and got the disk full error. I had to contact them and they fixed it. According to Statistics on the dashboard my current Disk Usage is 853.46 MB / 19.53 GB (4.27%) , Bandwidth is 305.99 MB / 488.28 GB (0.06%).

[NickMNS]

It is not you, there is a problem reported and discussed here:
[webmasterworld.com...]

[Dimitri]

Your root (/) partition is full. The disk is divided in several partitions, the whole disk might still have plenty of free space (for example in /home) but it looks like your root partition is full, which is a problem, the system has to right things in folders which are on this root partitions, and if it can't, then it can lead to this kind of problem. Check the size of the root partition (/) available space, available inodes too (sometimes too many small files, can fill all the available inodes, even if there is plenty of space). Check the /tmp folder (which is usually on the root partition), it's possible something has written too many temporary files there. Check also folder like /var.

edit 1: however, it might be more wise to let a specialist to check this, and clean the extra files. Do not try to delete things yourself, if you are unsure of what you are doing. You risk to break your system or make things worse.

edit 2: @NickMNS, the OP is not referring to GA

[DChan]

Thank you so much Dimitri, I am having the web host check into this because not know what I am doing I don't want to mess something up further. I was supposedly just moved to a new server on May 29th. They were having ongoing outages on the server I was on in May. On the 17th I received an email that they decided to move everyone immediately to a new server. I thought they had moved me but I was still having problem with sales even though I had the same amount if not more traffic. And because there was also a high bounce rate I decided to set up a monitoring service on May 28th just to see if there were still outages. I started receiving notification of outages by the next day. I contacted my web host, sent them screen shots of the monitoring details. They said they were just now moving me to the new server. That 95% of the people were moved but I wasn't one of them because I had a dedicated IP. Anyway, I am still getting outages notifications.

I just received a reply about asking them about the disk full problem. This is the reply, "Yes this is from migrations as we are still repairing the previous server in which we can then migrate some accounts from this new server to free up space. We haven't hit any limits though."

[Dimitri]

Might be the time to consider switching to a different webhost ...

[DChan]

Any suggestions?

[tangor]

@DChan ... we don't do "promotions" ... but you might want to look at a host that has "unlimited" ... there are a few out there, and some are surprisingly inexpensive.

[DChan]

Okay, thank you Tangor.

[martinibuster]

Paying a little more for a server can get you onto a server that isn't packed to the gills with tens of thousands of low quality sites sharing resources.

Also check around for a dedicated server. There are plenty of affordable options and if support is available then that can be super helpful.

I get the partition error once every couple of years on my dedicated server and I have a guy who fixes that for me when it happens. It's been a long while since that's happened though.

[lucy24]

"There are 0 500 errors shown in the access logs."

My immediate thought was: Oh, how sneaky, they’ve configured the server not to log 500-class errors. If you’re feeling paranoid, this is easy to test: Make up a rule that returns a manual 500 in response to suchandsuch request. Make the request from your own browser--noting the exact time!--and verify that you see a 500 in your browser. Now download and open the raw logs, and see if the 500 is there. (If you forgot to make a note of the time, just look for your own IP.)

503 responses also crop up when there are more than {some number} of simultaneous requests, where {some number} is set by the server administrator. This can be troublesome if the limit is something low, like 30 (the default is 100), if many of your pages have more than that number of supporting files.

[DChan]

Thanks to you all for your help. Yes, the trust is no longer there for this company! I have moved on! Now, when checking my server information at my new web host, the "Disk / (/)" column is at 1 %!

Thanks again, love this forum, so helpful!

[DChan]

"Make up a rule that returns a manual 500..." I am not sure how to do that and I have moved to a new web host but I would love to know how to do that!

[lucy24]

When you’re settled in, post in the appropriate subforum (Apache or IIS) and someone will walk you through it. If the new server is nginx we don't have a dedicated subforum; people tend to post those questions in Apache and hope for the best.

[mikejolley]

I would definitely check your Apache server logs to see what is happening.