No Response Errors after last update

B

bh-tech

Well-Known Member
#1
#1
We recently updated from 5.0.18 to 5.1.9 (legacy to stable)

Ever since the update was made, our Google Search Console has started reporting 'Server Connectivity' issues. In particular they are 'No response' issues.

The errors have only started appearing since the day I did the update, so hard to say they aren't related.

Out of the ~10,000 pages that Google is indexing each day, it is only reporting around 1% or 100 pages returning with 'No response'. So it's only having an issue for a very small amount of time. I can't get any other information from the Search Console report.

What could be causing this issue? Should I roll back to 5.0.18 to see if the issue continues?
 
B

bh-tech

Well-Known Member
#2
#2
This is the message from Google for this type of response:

Server error: No response

Google was able to connect to your server, but the connection was closed before the server sent any data.

Use Fetch as Google to check if Googlebot can currently crawl your site. If Fetch as Google returns the content of your homepage without problems, you can assume that Google is generally able to access your site properly.

It’s possible that your server is overloaded or misconfigured. If the problem persists, check with your hosting provider.​
 
N

NiteWave

Administrator
#3
#3
bh-tech said:
What could be causing this issue? Should I roll back to 5.0.18 to see if the issue continues?
Click to expand...
please force update 5.1.9 to latest build, there are a few new minor updates.
#/usr/local/lsws/admin/misc/lsup.sh -f -v 5.1.9

see if the issue persist.

even roll back, 5.0.19 is recommended.
 
P

Pong

Administrator
Staff member
#3
#3
You should check your server log for details. Any error message?

If google Search Console reports give you timestamp of "no response", you should try to look at log at matching timestamp.
I am not sure if Search Console reports give you reliable uptime report or not. Normally people use other monitoring tools, like Nagios. You can track when you receive any alert.
 
B

bh-tech

Well-Known Member
#4
#4
I run the script above.

It checks for new releases, downloads the latest build, states Memory size, license key created, then I get:
Require root permission to install this plugin. Abort!

What's going on?

I've run this command before to update and I'm logged in as a root user.
 
B

bh-tech

Well-Known Member
#4
#4
Pong said:
You should check your server log for details. Any error message?

If google Search Console reports give you timestamp of "no response", you should try to look at log at matching timestamp.
I am not sure if Search Console reports give you reliable uptime report or not. Normally people use other monitoring tools, like Nagios. You can track when you receive any alert.
Click to expand...
Hi Pong,

This is part of the issue, Search Console gives me nothing but the number of errors on each date. So I don't have any timestamps to go off. I've looked through our logs and found nothing that resembles these errors.

What would I even look for? How would a 'No Response' appear in a log?

I'm still getting the errors after updating to the latest build. It's just very strange that the errors started right after I updated to the stable build 5.1.9.

Is there anything else I can do?
 
B

bh-tech

Well-Known Member
#5
#5
Updating to the latest build has not fixed the No Response issue.

I'm getting these lines in my error_log which may be related:

Connection idle time: 61 while in state: 6 stream flag: 4,close!

What does this mean?
 
N

NiteWave

Administrator
#6
#6
bh-tech said:
Should I force reinstall from WHM?
Click to expand...
command line should be fine.

bh-tech said:
Require root permission to install this plugin. Abort!
Click to expand...
this is weird, since you already logged in as root.

it looks to be a linux permission issue, not because of any special of lsup.sh. maybe relate to selinux enabled ? I have never experienced such behavior -- run script as root user, but asking for root permission.
 
B

bh-tech

Well-Known Member
#6
#6
All these posts are out of order? My most recent post is up the page further - can you confirm my latest feedback has been seen?
 
N

NiteWave

Administrator
#7
#7
bh-tech said:
Updating to the latest build has not fixed the No Response issue.

I'm getting these lines in my error_log which may be related:

Connection idle time: 61 while in state: 6 stream flag: 4,close!

What does this mean?
Click to expand...
now let's back to the original issue.
https://support.google.com/webmasters/answer/35120?hl=en
"Use Fetch as Google to check if Googlebot can currently crawl your site. If Fetch as Google returns the content of your homepage without problems, you can assume that Googlebot is generally able to access your site properly."

how about "Fetch as Google" result ?

bh-tech said:
Connection idle time: 61 while in state: 6 stream flag: 4,close!
Click to expand...
this may relate to setting
lsws web console -> Server -> Tuning -> Connection Timeout (secs)
default is 300 seconds. have you changed it to 60 seconds ?
 
N

NiteWave

Administrator
#8
#8
no, reinstall has same result, either via command line or WHM.

what's the build version ? the info is in /usr/local/lsws/autoupdate/build
 
B

bh-tech

Well-Known Member
#8
#8
NiteWave said:
how about "Fetch as Google" result ?
Click to expand...
Fetch as Google returns everything fine. The issue is happening 1% of pages crawled - which equates to around 100 pages per day. I don't know whether this is 100 in a row, or 100 sporadic no response errors.

NiteWave said:
this may relate to setting
lsws web console -> Server -> Tuning -> Connection Timeout (secs)
default is 300 seconds. have you changed it to 60 seconds ?
Click to expand...
Yes this is set to 60 seconds - though I'm not sure when it was changed - should I revert to 300?
 
B

bh-tech

Well-Known Member
#9
#9
I've just looked back through our logs and seen that before we updated to 5.1.9, (we were on 5.0.18) the Connection Timeout error reported as this:
Connection idle time: 61 while in state: 2 watching for event: 25,close!
and since updating to 5.1.9, the error looks like this:
Connection idle time: 61 while in state: 2 stream flag: 4,close!

Not sure if that means anything, or if it's just a difference in the way the versions report the same error - but though I'd include it.

I'll also add that these connection errors have been happening before Search Console started reporting the No response issues - so that probably means the two are not related.
 
N

NiteWave

Administrator
#10
#10
bh-tech said:
Fetch as Google returns everything fine. The issue is happening 1% of pages crawled - which equates to around 100 pages per day. I don't know whether this is 100 in a row, or 100 sporadic no response errors.

Yes this is set to 60 seconds - though I'm not sure when it was changed - should I revert to 300?
Click to expand...
yes, please revert to 300 and monitor again.

please get a list of all such changes which is different to default value --- most default value is not recommended to change unless it's very clear which need to change it.
 
B

bh-tech

Well-Known Member
#11
#11
NiteWave said:
yes, please revert to 300 and monitor again.

please get a list of all such changes which is different to default value --- most default value is not recommended to change unless it's very clear which need to change it.
Click to expand...
Where can I find the default values for other settings?
 
N

NiteWave

Administrator
#12
#12
you can install a fresh lsws on another test box and check with /usr/local/lsws/conf/httpd_config.xml (may need get a trial license)

or download a latest package, extract the .tar.gz, check conf/httpd_config.xml.in
 
B

bh-tech

Well-Known Member
#13
#13
I've found what I believe to be the defaults.
Here are our differences:

Tuning | Our Server | Default
Connection Timeout | 60 | 300
Max Keep-Alive Requests | 120 | 1000
Keep Alive Timeout | 3 | 5

I will adjust the Connection Timeout - should the others be adjusted?
 
B

bh-tech

Well-Known Member
#15
#15
Our Max Keep Alive Requests is quite a lot lower than the default - could that affect people connecting to the site during busy times?
 
N

NiteWave

Administrator
#16
#16
1) suggest just change back "Connection Timeout" to default and watch/monitor "Server error: No response" issue

2) if the issue disappear or improved a lot, revert back other 2 parameters as well and watch again

3) if this issue not occur again or at least no worse/no better, keep all the 3 default settings.

per my experience of mine and quite some lsws users, tuning such default settings may take a lot of time, and finally found it's just a waste of time when looking back.

so my advice has been there for a long time: don't try to tune most of default settings in lsws unless you're very clear what you're changing.
 
B

bh-tech

Well-Known Member
#17
#17
1) I have adjusted the Connection Timeout to 300 sec. In reference to the 'Connection Idle Time' error that I reported, I am still getting them every now and then - only difference being the '61 sec' is replaced with '301 sec'

2) The report from Search Console is behind by around 2 days - so I will be able to tell in 2 days if the error has changed.

I also updated to 5.1.10 this morning, so that could potentially help too.

I will report back here when I have more information.
 
B

bh-tech

Well-Known Member
#18
#18
So Search Console is reporting the issue as OK - however looking back at the reports, the issue actually got better two days before I made those changes above - which makes zero sense.

I will keep monitoring to see if anything else changes - but it's very strange as the change to the Connection Timeout doesn't correlate with the reports.
 
B

bh-tech

Well-Known Member
#19
#19
UPDATE

The issue seemed to have fixed itself for like 10 days, and how it's happening again.

Should I look at reverting those other two parameters to their defaults?
 
You must log in or register to reply here.
Top