Rubénerd!

I just finished the transition from SegPub to Net Logistics, as you can tell from my terribly sophisticated line-up of icons above!

What colour is whoa?

I had my new account activated in less than an hour after I made my first monthly payment. In less than 30 minutes I had my private/public keys for SSH/SFTP swapped and usable, six years of blog posts restored in a new database, six years of images uploaded through SFTP, WordPress uploaded and installed and all my custom .htaccess crap implemented.

Good heavens, the difference in performance is amazing, its like someone shoved rockets up my blog and bidness card sites' arses. I could have phrased that better.

This heading is also entirely pointless

I get the feeling I'm going to like it here. Thanks for all your help Martin, Tarinder and Alicia A. Automation. They're so cute, they even name their bots. And they operate in Australia and Singapore... just like me. *sniffles*

If something isn't working, leave a comment here and I'll patch it up. In return, I might start blogging about things that have nothing to do with web hosts again.

Who knows, I might be able to host my assignments reliably this semester so I don't get failed for them!

SegPub really outdid themselves this evening; my sites were (and still are) on average taking six minutes to load, and that was only when their domains decided to resolve at all.

I give up SegPub, I'm moving. You don't answer your support tickets, messages I send to your mailing lists are rejected and your service is getting laughably bad. My GeoCities page on dialup was faster.

I'm just glad I don't have any assignments for the next few weeks. SegPub have already cost me one high distinction when a demo site failed to load and I was failed for that part of it. Thanks guys :(

As my old man would say... bummaire

The Problem

This morning Rubénerd suffered yet another SegPub problem, this time the software couldn't connect to the database not because of an authenticate error but because it claimed it didn't exist. After roughly 15 minutes, the entire site went down.

Everything seems to be back to normal again as of 12:39 Sydney time. Including this one, this is the eighth SegPub problem I've documented since 2009.

I tell you what, RootBSD.net is looking better and better. I need a cup of coffee.

The Update

I just received an official communiqué from SegPub this afternoon. I like the word communiqué, it makes me sound cultured. Does that mean I'm a yogurt?

Hi Ruben,

Apologies for the delay in getting back to you. Earlier today we suffered an outage in one of our internal DNS caches, that caused some internal DNS resolution issues, including being able to connect to our database cluster. The issues were fixed once we were aware of the problem and you shouldn't be seeing any more database problems.

I appreciated their timely response, but I still would have appreciated a heads up email when they knew something was going wrong instead of having to ask. Oh well, life goes on.

In late April, my webhost SegPub informed me the specific webserver causing problems for my site and the sites of many other people was in the process of being fixed. It's now almost June and the problems are still persisiting for all of us, if the newsgroups are anything to go by.

Some small comments:

"Could you *please* keep us in the loop as things progress. The silence on this issue has added to the misery of having sites that are so often unreachable (certainly more than 0.5% of the month)."

"I have 2 site on Chewie and have had endless problems, not the least being NO response to any support ticket ( I've lost count how many I have sent). Hopefully things improve sooner rather than later."

"Alas, the problems seem to be persisting. Once again, [my site] is down, while all of my other SegPub-hosted sites are up."

"Same here, sites constantly go-slow or time out."

Peter wrote the best summary:

"Yes, I'm still seeing persistent slow response from [our server] too. This appears to have started around February 2010.

I opened a support ticket a month ago, sent a reminder after two weeks without a human response, then received the answer "In regards to the speed issues, we were having some issues with bots flooding [our server] with traffic at sporadic times of the day. We believe we've resolved this and you should be seeing the site load at acceptable speeds."

I sent copies of my monitor reports, which show that performance is till poor when compared with a site on another host's shared server.

I have just received a monitor report that the site on [the server] was down for three hours.

Altogether disappointing."

And as another cruel twist, this post has taken an hour to go live because the site keeps timing out whenever I hit the Publish button. I've decided to give them a couple more weeks before I start looking for alternatives to their service.

RootBSD look like a great FreeBSD VPS service, or for the greenie hippie in me there's always Green Geeks that Marco recommended here a while ago :). I tend to get a bit nervous with sites that oversell though, but I'll check them out.

I'm not sure whether these problems are just the typical WordPress fare or because SegPub is playing up again. Either way, I can't wait to get a new host and start fresh with some nice Django goodness, I'm getting so tired of all these problems :'(.

Finally some good news from Segpub regarding the timeouts and slow response times (3 minutes, 38 seconds for a plain HTML file?!) that have been sporadically plaguing my poor sites for months. From a mailing list email addressed to folks on my server:

The issues with [your server] unfortunately are related to search bots and a mix of other nasty traffic. We're working really hard to get ontop [sic] of these and are pretty close to having everything resolved and back to normal.

Nasty traffic? That's not a nice thing to call my Nonsense section! ;).

The problem is, because my sites slowed down to a crawl but still eventually loaded, they don't count these problems as downtime which means they won't be providing me any free months as stipulated in their uptime guarantee. Sneaky.

This morning's intermittent Rubénerd outages were brought to you by Segment Publishing. They're still miles better than Servage was, though that isn't saying much.

Curiously, many of the outages over the last year have been around this time of day. I wonder if they do server maintenance around 00:00-02:00 Adelaide time, or whether it's due to increased loads given it's daylight in the US.

UPDATE: Іt's now 10:42 and pages are still taking many long minutes to render. Time to move web hosts again?

Some more of SegPub's legendary reliability from earlier this evening. I suppose it'd been a whole month since my last outage ;).

Another SegPub outage this afternoon, this time lasting from around 14:32 to 14:52 South Australia time. Fortunately I wasn't using the domain for any of my afternoon classes like the last outage did.

This time I did a traceroute and checked the status of the site using a foreign server and recorded the results. I'm hoping this isn't becoming a regular occurrence, but I'm going to be keeping my own records of such events.

Today has not been a very good one for the intertubes. For the second time in the last couple of days Rubenerd.com went offline for an extended period of time, fourth time if you count database connection errors. SegPub guarantees a free month's hosting if uptime dips below 99.5%.

The timing could not have been worse, some of the blog posts I had written regarding OpenSolaris and Java were supposed to be used in a practical today because I thought I could get some valuable feedback (which I did) instead of just writing it up in OpenOffice or Word. I showed the results of a well known downtime indicator to the teacher, but he was not impressed. Serves me right for trying to think outside the box.

I got a reply from the SegPub discussion mailing list regarding the matter:

I can assure you we haven't had any outages recorded today. We monitor all our services from 19 redundant locations, if 2 or more report an issue then we consider a service to be down and we'll react immediately.

I'm not saying it wasn't down for you, but just in regards to our monitoring, we haven't had any outages today. To give you a bit of background, this is the first time since February 2009 that we've had any kind of issues, its just unfortunate when things happen they tend to happen all at once.

It'd be great to see if we can get some source IP addresses you were connecting from so we can trace the issues at our edge routers so we can trace back and work out if perhaps there's a network issue along the way.

I can assure you we take downtime VERY seriously, and we strive to maintain our excellent uptime and stability.

I believe them when they say they take downtime seriously, but respectively their claim that there was no downtime today is simply not true.

I could understand if it was just my isolated IP address or ISP having connection problems, but I was able to confirm with people living in Singapore and London that it was offline, and as you can see from the screenshot from the following website they were reporting it down too.

I've had nothing but positive experiences to report with SegPub, they run a tight ship. Here's hoping these last couple of days were just one off occurrences. If it does happen again I'll be running a dtrace to see where the problem starts.