Post by: krissel on October 31, 2008, 06:18:59 PM
The site may be unreachable for a few hours so don't worry, that is if you are awake.
Post by: Paddy on October 31, 2008, 06:44:28 PM
We're hoping that this will solve some of the problems we've been experiencing lately, which we think are due to an overloaded server.
Post by: Paddy on October 31, 2008, 08:23:19 PM
Didn't get much info back re: server load - the response was:
QUOTE
The server load was a little high, but nothing outrageous. There wasn't much that could be done to lower it though.
Okaaay. Let's hope that whereever they move us it's better!!
Post by: krissel on November 01, 2008, 01:54:29 AM
Post by: Highmac on November 01, 2008, 04:09:48 AM
Post by: sandbox on November 01, 2008, 07:41:11 AM
Post by: gunug on November 01, 2008, 07:49:50 AM
Post by: Xairbusdriver on November 01, 2008, 10:41:51 AM
Post by: Paddy on November 01, 2008, 11:36:45 AM
SB, I don't know what server they're moving us to - I have asked as part of the whole request, but got no response. I will of course ask again when it comes time to actually reopen the ticket, and I will also ask to be given a few minutes to check out its load etc. BEFORE they move us. I was told us that the server load on Analucia was "high but not outrageous, but they couldn't do anything to lower it" - um, methinks a server load of 10.5 or so on a 2-CPU server is a mite outrageous. That happened yestereday morning. Last night it was 3.84 when I checked - STILL too high. It should be less than 2. And the Apache Server status was reporting up to 40% at times, which was higher than I'd seen on any other ASO server. The weird thing is - I just checked it now and it's down to 3.69% in the Apache Server Status and .88% in the Cpanel server status report - I've never seen it so low. I'm now wondering if our complaining somehow had some effect?
SB - can you tell me the difference between what is reported in the Apache Server Status ( http://analucia.asmallorange.com/server-status ) and what we see in the CPanel re: CPU load? It's something we've been puzzling over and trying to figure out - and trying to figure out what figures we should be concerned about.
Post by: Paddy on November 01, 2008, 10:31:24 PM
Today I looked at our server stats several times - the CPU load is now considerably LESS than some other servers at ASO; in fact the loads have dropped a lot on many of the servers. Curiouser and curiouser.
So, having a poke around again and seeing that the numbers had stayed low, I emailed ASO tech support and asked why the sudden change? And here's the response:
QUOTE
Hello,
I have looked into this for you and it appears that our system admins have performed an audit on this server and found a few users that were causing some higher then normal load but this abuse was not showing up in our process slight. It looks like they finally tracked down the abuses and have stopped them or eliminated them from the server. Our Admins are always watching the servers and we do our best to catch those causing issues for our Shared server clients.
Please let us know if you notice any more of these previous server issues as we will continue to improve our services for all of our clients.
Please let us know if we can further assist you with anything else.
Thanks,
Ryan
I have looked into this for you and it appears that our system admins have performed an audit on this server and found a few users that were causing some higher then normal load but this abuse was not showing up in our process slight. It looks like they finally tracked down the abuses and have stopped them or eliminated them from the server. Our Admins are always watching the servers and we do our best to catch those causing issues for our Shared server clients.
Please let us know if you notice any more of these previous server issues as we will continue to improve our services for all of our clients.
Please let us know if we can further assist you with anything else.
Thanks,
Ryan
So it would appear that the squeeky wheel did (finally) get some grease! I suspect they may have had a good hard look at all the servers as a result. The highest load I could find in a brief perusal (A through E) was about 20% - much better than any stats I'd seen in the past few days. Analucia is less than 1%.
So, I think we'll sit tight and see if things improve, rather than move right away. There may be no need now.
Post by: sandbox on November 02, 2008, 01:15:20 AM
Now that they discovered the culprit and if the capacity is not too high you may want to stay put. I'm sure you've considered this as well but in my experience I've found that they will treat the wounded or abused server with a little more attention in the future and not burden it with a heavy load.
Post by: krissel on November 02, 2008, 01:57:18 AM
Please keep posting to alert us to any time you haven't been able to get to TS.
Mark, I fixed the link. It doesn't have the www.
Post by: sandbox on November 02, 2008, 11:56:58 AM
QUOTE
SB - can you tell me the difference between what is reported in the Apache Server Status ( http://analucia.asmallorange.com/server-status ) and what we see in the CPanel re: CPU load? It's something we've been puzzling over and trying to figure out - and trying to figure out what figures we should be concerned about.
Paddy, what I've always just took for granted, whether it's correct on not, is that the load range on cpanel was an average of the timeline indicators ?,1,5,10,30,? minute spread and the status page defines the specifics for the techs. I have used cpanel load range as a reference many times when writing tickets and they were mostly answered with a reference to the increase or decrease in the load indicator. I do recall once though, that my ticket seemed like it was behind the curve and that the issue was a spike rather than a trend but the indicator showed a high load over a long time period, which would indicate an average if the problem was resolved. I'm just assessing this from reconstituted memory so I'm not quite sure the answer is accurate.
Post by: Mrious_be on November 03, 2008, 02:37:51 AM
Post by: Paddy on November 03, 2008, 09:57:25 AM
Post by: Xairbusdriver on November 03, 2008, 01:58:01 PM
Post by: Jack W on November 03, 2008, 06:27:30 PM
- Jack
Post by: Paddy on November 03, 2008, 07:16:34 PM
We will continue to monitor this situation of course - please continue to post any outages you see and check the server status when it comes back up, particularly if it was down for more than a minute or two, so we can document the length of time of the outages (approximately).There is one user in particular that seems to be driving things nuts (really long times to process requests) and I'm going to file a ticket on that to see if there is still some funny stuff going on.
Post by: krissel on November 04, 2008, 02:00:41 AM
http://analucia.asmallorange.com/server-status
This one lies...
http://analucia.asmallorange.com/uptime.php
Server http was down tonight for about 25 minutes around 1:15 AM so the time on the first link is correct. The second is blatantly false unless the first notes any failure and the second records when all three aspects of the server go down. Just don't know.
Anyway, server load shortly after it came back up was 19.35 out of 2 cpus.
Just now it is .63.
So there is some pretty serious spiking going on.
Post by: Paddy on November 04, 2008, 10:43:47 AM
Apparently, Apache Server Status reports on Apache - not on whether the server itself is up or down. Of course, for our purposes, it's one and the same, since without Apache, we're dead in the water. Ditto the CPU load reporting in that status report - it's reporting on what percentage of the CPU horsepower is being used by Apache. Obviously, the higher it is, the higher the load, though if something is busy making all sorts of calls via MySql etc. the total CPU load could be high even if Apache is low(er).
The CPanel server load was 19.35?? (or the uptime.php link - same thing) If that's the case, that's insane and I need to add it to my ongoing conversation with Tech Support (I'm now talking to a Level 2 tech).
I'm fairly sure of where I want to migrate to: http://rose.asmallorange.com/uptime.php (it's the Northlea server) or maybe this one: http://isabel.asmallorange.com/uptime.php
Post by: krissel on November 04, 2008, 01:39:35 PM
[attachment=1101:server_l..._11.4.08.jpg]
Out to vote and to the folks... later.
Post by: Paddy on November 04, 2008, 02:35:10 PM
QUOTE
Hello,
We've been looking into this and have rearranged a few accounts within our servers to balance the load better. This should resolve the long term issues you have been seeing lately.
Regards,
Dennis
We've been looking into this and have rearranged a few accounts within our servers to balance the load better. This should resolve the long term issues you have been seeing lately.
Regards,
Dennis
Let's hope it does resolve the problems...we will be watching!
Post by: Xairbusdriver on November 04, 2008, 04:05:38 PM
Post by: sandbox on November 05, 2008, 03:08:32 AM
Paddy try to get to Tim, he runs the place and should advise you on your best course of action. If you have any trouble drop me an email and I'll talk to him.
Post by: chriskleeman on November 05, 2008, 07:57:43 PM
The server on my website, "Karen" was having her own set of troubles tonight earlier around 7-7:45 P.M. In and out, up and down...
Seems to be cleared up now...
Chris K
Post by: Xairbusdriver on November 06, 2008, 09:25:52 AM
Post by: RHPConsult on November 07, 2008, 03:14:38 AM
Post by: krissel on November 07, 2008, 04:28:41 AM