Jump to content

Recommended Posts

Posted

Speeds were better at night, though still plenty of timeouts to be had.

 

As of right now (10am PST), speeds aren't horrible. Timeouts/half-loaded pages still occurring, but the site is navigable, which is a great improvement over the rest of the weekend. Average time of page load the last 12 hours is between 3-5 seconds, whereas this weekend was 5-10 seconds.

 

Will keep an eye on it and report if anything changes.

 

EDIT: (3pm PST) FTP/cPanel are pretty much useless at this point. Timeouts ad nauseum. Site is still accessible, though very inconsistent. There is definitely a correlation between time of day and server speed, which probably has to do with increased activity.

 

I really can't say what's up. If all that has been done hasn't restored Stevie, my intuition says hardware related (which is a possibility since these changes occurred right after a power outage).

 

Color me stumped. :mellow:

 

EDIT 2: I've made this little picture from the site monitor that was posted earlier. Basically you can see that Stevie looks to be much slower now than it was earlier this month immediately prior to the outage. When the graphs become green, site speeds return to normal, but you'll notice that green correlates with the times of low activity. The possibility exists that this is imply due to increased users on Stevie, though the timing suggest relation to the power outage.

 

  • Replies 127
  • Created
  • Last Reply

Top Posters In This Topic

Posted

Never mind... after a further look, my theory doesn't make sense.

 

At this point, I really don't know what to say. There's tons of memory available (can't be a RAM issue), CPU load at a reasonable level (can't be a CPU issue), and disk I/O is stable (can't be a disk issue). What could the issue be?

Posted

Something I've noticed since the power outage is that the server load has been extremly low ALL the time. Most of the time under 6 or 7. It's never been that low since we had all of the users we have.

 

http://byrondallas.heliohost.org/server_load.php

 

Does that tell you anything? Could it be it's not returning the correct server load?

 

 

 

Guest Geoff
Posted
  Quote
is it Worth to reboot again?

 

If disk IO is stable, there is no point.

 

@djbob Maybe there is a thread/process limit or something that is causing apache to Queue requests.

Posted

byron: I think that's probably because the server only has 1/3 as many accounts at this point...

 

Geoff: There is a limit, but if it was being reached then PHP would just die and return a 500 error instead of timing out.

Posted
  djbob said:
byron: I think that's probably because the server only has 1/3 as many accounts at this point...

 

I hope your right because I'm happy with the server loads now but check out the loads right before the outage which was May 6th.

 

http://byrondallas.heliohost.org/server_load_may.php?page=3

 

They were a little higher than they are now. Seems we really stay remarkably low now. Regardless, it may not have anything to do with it, but thought it worth mentioning. :)

 

 

 

Guest Geoff
Posted

Some of this could be time spent on looking up NS records. (Which isn't even on stevie)

Posted
  djbob said:
Geoff: There is a limit, but if it was being reached then PHP would just die and return a 500 error instead of timing out.

I often get a lot of 500 errors when trying to load my PHP pages (approx. 8-10PM PST). Then I have to wait a few minutes before refreshing the page or it gives me the error again. So some limit must be being reached. But the server load has been weird. Before the power failure, Stevie's server load usually stayed around 2-3 or would jump up to 14+. Now it usually hangs around 6.

Posted
  Geoff said:
Some of this could be time spent on looking up NS records. (Which isn't even on stevie)

 

 

In a case of my monitoring server - I send the requests directly to the server with using IP address, so any DNS issues does not affect my results. What's more - I see that majority (if not all) "A" records on heliohost nameservers have TTL set to 4 hours, so in such case every person would have one slow query per 4 hours only for selected domain name, later DNS entries are cached on his/her computer. Additionally - as you may see just now for hours - while Johnny does not work, Cody is very fast, so the responces from nameserver, and situation on Stevie does not change.

 

 

Posted
  iceitsup said:
  djbob said:
Geoff: There is a limit, but if it was being reached then PHP would just die and return a 500 error instead of timing out.

I often get a lot of 500 errors when trying to load my PHP pages (approx. 8-10PM PST). Then I have to wait a few minutes before refreshing the page or it gives me the error again. So some limit must be being reached. But the server load has been weird. Before the power failure, Stevie's server load usually stayed around 2-3 or would jump up to 14+. Now it usually hangs around 6.

 

well i get this issue too but there is no time related about it .

i think that there is a limit on connections per site , so if you upload some files to your public_html

the numbers of connection is raised up .

i am getting this 500 error also . i don't know how to fix it maybe staff members will help about this issue.

 

to the issue related server load as shown on monitors its really weird but i don't know how its calculate it .

main issue is that might be some background process like cron or some other job that take all power to him self or hardware related problem.

 

the corrective disk supposed to be at list 24 hours but it has done in 2.5 hours & no errors found so i believe that it didn't evan worked right .

some error must be found cuz the power failure shut down the server on spot & no close file had been generated .

when you start the server after power failure it need to show you some log about missing file or message about power failure or shut down unsuspected

maybe in this log you got your answer to the issue .

 

EDIT: sorry about my English it isn't my main language

Guest Geoff
Posted

If you get a 500, you have exceeded your process limit.

Posted
  djbob said:
Never mind... after a further look, my theory doesn't make sense.

 

At this point, I really don't know what to say. There's tons of memory available (can't be a RAM issue), CPU load at a reasonable level (can't be a CPU issue), and disk I/O is stable (can't be a disk issue). What could the issue be?

There's enough RAM, but could it be that the sudden shutdown physically damaged it, or just messed up being able to read and write to it?

Posted
  SomeKev said:
There's enough RAM, but could it be that the sudden shutdown physically damaged it, or just messed up being able to read and write to it?

 

 

good point

Guest
This topic is now closed to further replies.

×
×
  • Create New...