Longstanding issues

[Ron]

Ok, here are a couple of longstanding issues I've had with Jag...

1) When support (which has been doing really well!) responds to an issue, the response almost always reads something like this: "The site is now working fine. If you still have a problem let me know"

I don't know if this means that they were able to re-create the problem and have corrected it, or if they were unable to recreate the problem. I find this abiguity to be frustrating.


For instance, today I had a recurring problem: Neon HTTP *ONLY* is slow as hell. https is fine, SSH and FTP are fine, but HTTP is horrendous. So I re-open a ticket from when this was really bad 11 days ago. This is what the ticket said:

+ + + Client + + + : 2004-02-12 / 17:55:40
Hello,

HTTP slow on NEON. HTTPS is fine, SSH is fine.

Thanks.


- - - Tech - - - : Support Tech 2004-02-12 / 18:16:19
Hi

Please check it out now. Its working fine now.

Regards,
Suresh
Support Dept.


+ + + Client + + + : 2004-02-23 / 16:51:31
Hi folks,
I am replying to this ticket instead of opening a new one. This has been a recurring problem, that is showing up again. Can you help?

Thanks!
-Ron.


- - - Tech - - - : Support Tech 2004-02-23 / 18:59:57
Hi,

I checked the server Neon. Everything is working normal. The sever load is also normal. I have pasted the results below. Are you still experiencing the problem. Please let me know.

--
root@neon [~]# uptime
7:56pm up 14 days, 22 min, 3 users, load average: 0.47, 0.93, 1.56
root@neon [~]#
--

Regards,
Prasad,
Support Dept.

Well, things cleared up about 10 minutes before I got this email. Did support fix something? Yes? No? Not sure.

So then I remember... back in Aletia days, I had set up a watchdog to monitor my site.... Are they still generating reports? Yes! So here is my reply:

+ + + Client + + + : 2004-02-23 / 20:00:32
The problem cleared up about 10 minutes ago (about 1 1/2 or 2 hours after the ticket I submitted. Prasad, this has been happening fairly regularly, and eventually support had to rebott NEOn to fix the problem. I know that Jaguar monitors the server and HTTP, but can they also be monitored for SLOW response time?

I just thought of something, hold on...
I used to have to monitor my site all the time when it was with Aletia... I got a watchdog service then, and I just decided to look up my report -- and they are STILL monitoring my site! Cool. Anyway, take a look at the following graph for response times to my site:
TODAY:
http://www.oes.org/temp/WDgraph.gif
Last 7 Days:
http://www.oes.org/temp/WDgraph-7.gif

Notice the long wait times today, and over the last 7 days.

Thanks for you help!
-Ron

Now, in my logs, I can see that they looked at the images (from India using Opera under Linux) about 20 minutes later...

I still haven't heard back from support. Maybe I wasn't specific enough in my response. Maybe I should have written: "Hey! My site was getting 28 second response times, what's going on? What are you going to do to ensure that I don't have to be the one to notice that my site is unavailable, and that after I notice it that I don't have to wait for 2 hours to have my site back up again??"

Is there something going on with Neon? Was it (if anything) fixed? Will this happen again and again, until they need to reboot again?

More communications would be really nice.

2) I am STILL (after complaining about this since the merge of the forums) not getting email notifications from this bulletin board.

Other than that, I was very happy to pay my $100 to Jag to host my other dormant site that sees about 5MB traffic per month. I was going to multihost it to another account, but I was lazy and figured Jag has been so good.... Then a few days later, I get whacked. Timing is everything.

[Ron]

Ok, so now I get the following response:

- - Tech - - - : Support Tech 2004-02-24 / 00:05:13
Hello Ron,

I went through the graph. We did not have any apache issues. If the response time is low, the server load would be high and we would be getting alerts about it. Or else, it should be network problems. So when you are noticing slow response again, in the very second, please do a traceroute to the domain and follow-up to this ticket. We shall try our level best to sort it out.

Regards,
Bindiya
Support Dept.

Ok, valiant lip service. The point is, the system had slow response times both from my home in Massachusetts, as well as my watchdog system run by Global Crossing based out of Ohio. Now I'm told again that *I* have to be the one to discover slow response times. Since I have found slow response times on several occasions, and it took two hours for support to respond to my ticket, what else can I do?

They want me to send a traceroute -- all well and good, but if the problem is with HTTP traffic only what good will that do?

Last time this happened, less than 2 weeks ago, I don't know what the problem was, but I got the "everything looks good" email. Back in December when this exact same thing was happening, Prassad said there were many processes listening to port 80, and he restarted the web server, and everything cleared up. Having something crowd out port 80 would cause there to be long response times while having an unusually low server load.

If it's taking 10-20 seconds to respond (to whatever program Jag is using to monitor HTTP), obviously that should raise an alarm, which it apparently does not.

4 straight hours, and nobody at Jag support knew about it!!!

Maybe I can set my watchdog program to send an IM to Jag anytime response time is > 10 seconds. Then again, couldn't they just do the same thing?

I was taught many years ago that 8 seconds was the "natural" limit for someone to wait for a response from a terminal. After 8 seconds people assumed that the machine was down, and/or got frustrated. I don't know what they're teaching nowadays, but I'd bet that attention spans are shorter, and people using my forums are going away in disgust when the system responds so slowly for 4 hours.

C'mon folks. Please, monitor HTTP for reasonable response time as well as failure. Please?

[lijoe]

Hi Ron,

The basic problem seems to stem from the fact that when all the services on the server are up and the load on the server is normal, like anyone else the techs too seem to accept that there is no problems.

Checking on the server indicated a speedy server resposne to http requests and hence for proof a few friends from here and there checked the site on their browsers to verify the same and found it acceptable.

We just checked the server right now too and there were a few backup processes running.This could have possibly caused you to get a slower access and we may have overlooked that factor.

Being the support to clients, we have to answer to a wide variety of clients and we assume most of the clients would be happy with a solved answer than being taken to the intricate details of the cause.

Anyhow we have checked it for now and find everything normal.Next time this happens, please open a new ticket so that we can promptly check what is causing the slow access.I request you to kindly check using your site IP address too before you come to us , so that we can know if it is a name resolution issue or a network issue also.

We hope to ensure that we will keep in mind that you are a client who would understand the details if said and shall make sure we will post you details rather than the solved response .

[Ron]

I'll be glad to do that. The possibility about DNS issues escaped me -- although I'm not sure how that could cause http: requests to be delayed while httpS, FTP and SSH requests to sail right through.
Thanks for your follow-up.