Website problems.

[whole diary]

[devon_metro]

In a strange twist of events, one of the only websites I can physically access is this 

Not even Google works and somehow I've mucked up my Wireless. Stay online or i'll loose my will to live 

[grahame]

We may be spluttering around 0.5% of the time at the moment, but we ain't going away 

[swlines]

We may be spluttering around 0.5% of the time at the moment, but we ain't going away 

I reckon it's more than 10% of the time according to my monitoring.

[Shazz]

We may be spluttering around 0.5% of the time at the moment, but we ain't going away 

I reckon it's more than 10% of the time according to my monitoring.

Only another 2% to go, and you've caught up with fgw

[grahame]

I can now give you a techincal background to why my reports were around 0.5% downtime but others were much higher. Basically, our heartbeat script that took the server off for a couple of seconds (if necessary) while it cleared up was running at the top of the minute. And so was the automated script that one of our users was running to see if we were up or down.  So (and probably unintentionally) he was always seeing the bad, and never the good.

I've blogged a more detailed explanantion with what I believe to be close to accurate figures that each of us was seeing as an example.

[swlines]

Ah but Graham, if it was indeed 0.5% then a lot of us were visiting at entirely the wrong times. After all, your heartbeat script did sometimes entirely collapse and lead to downtimes of 2 hours plus.

There are lies, damned lies and statistics - but having a script that monitors uptime based on when it can actually reach a webpage (ie, what the end-user can see as opposed to what your server is supposedly providing) is a better measure of uptime/downtime to me!

Anyway, I'm pleased to say that I've noticed a big difference in availability of the forum as has my scripts (hey, I'm not afraid to say it was me checking uptime!) - for the four days I've been checking uptime when the reliability was dodgy, to say the least, uptime stood at 84.6%. Outage reasons included response timeout, connection timeout, cannot connect to port 80, 404 Not Found, 500 Internal Server Error, etc. Since then it's all worked fine and I've not had one record of downtime except an erroneous fault this morning I suspect with a network as not both monitored it. For what it's worth, a couple of seconds each minute is about 4/60th ... and the script purposely randomised the time it polled the server - so I think your figures are a bit skewed on that one if you are honestly saying it was only 0.5%!

Shame it actually had to come to people really moaning to mods about how frequently it was down, and now the significant phrase, for the users. After all, this forum is meant to be for the people, so it's normally wise for it to be online for the people .... 

[John R]

Yes, but don't forget Graham does this all in his own time and funds this service, so I don't think we should be too critical that he has had some problems of late. He does have a business to run, and seems to have worked hard in his spare time to resolve what don't appear to have been simple problems.

So let's not be too critical?

[Andy W]

Yes, but don't forget Graham does this all in his own time and funds this service, so I don't think we should be too critical that he has had some problems of late. He does have a business to run, and seems to have worked hard in his spare time to resolve what don't appear to have been simple problems.

So let's not be too critical?

Hear, hear

[swlines]

I'm well aware that this site is run in Graham's "own spare time" but I'm now going to be extremely picky.

It could be said that this forum is used as a sales ground for Graham - as for some time the bottom of the page contained a link to www.phpcourse.co.uk and I don't believe it was removed until I picked up on it and told a certain moderator about my concerns to the link being shown on a forum. It could also be construed that the link to Well House Consultants (which was my other issue) in the footer could be also used for drumming up sales for the business. While I don't disagree with showing who hosts a forum, I feel on a forum that is named directly after FGW^▸ (First Great Western) it could be construed as if FGW personally condone the hosts (which I highly doubt they do).

Over a period of 4 days, the site was down for over 10% of the time - and I only really started monitoring after the downtime started to become unbearable (it is extremely frustrating for the forum to go down for, and I now quote, "for a couple of seconds at the top of every minute". When using the internet it is nearly impossible to enforce your browser to not contact a server for say, the first 10 seconds of every minute (it can be coded, but what time do you use? Your computer's time could be entirely different to the servers for instance).

In addition, I'm surprised that Graham didn't take any action when it became apparent there were problems, sooner, because savethetrain.org.uk, Melksham Rail Development Group, etc, etc, were all affected. After all, it doesn't reflect well if you're campaigning and your website randomly goes down every other 15 minutes. I'm sorry if any of you feel this is critical, but I have done a lot of work in this general area of industry (with success and failure in various attempts) and I know what I'm trying to say (and I hope Graham sees it as well, in the light I am aiming to show it in).

[John R]

I feel on a forum that is named directly after FGW^▸ (First Great Western) it could be construed as if FGW personally condone the hosts (which I highly doubt they do).

I think there's a very clear disclaimer at the bottom which disassociates this site from FGW's official site, and moreover provides a link to their site. So I would dispute that point.

[swlines]

You mean "This forum is provided by a customer of First Great Western, and the views expressed are those of the individual posters concerned."?

Seeing as "Hosted by blah" is right at the end of that same disclaimer (and it not being a view, at all) opens that disclaimer up for debate on that subject.

[Btline]

No, John R is right.

The website makes it VERY clear that it is not connected with FGW^▸ (First Great Western).

[grahame]

Gentlemen - Andy, John  and BtLine - many thanks for your support. From his latest post, there are clearly many things that swlines (Tom Cairns) is not aware of concerning all he work that his been going on in the background - and that's fair enough considering I had to keep a lot of the work below the radar, knowing that we had strange patterns of access going on ... some of which turned out to be Tom himself!!!

Some extra points / clarifiactions:

1. The problem was a "real doozie" to sort out.  Work was ongoing for some considerable time, and action was taken long before you realise - in fact action started the first time we were 'hit'.

2. Until I had the problem tracked down (researched) I was unable to tell you what the outcome of that research would be, or even when it would be concluded. Just like a soccer team can't tell you how long into the season it will be until they paly a game in which they score 3 goals.

3. While investigating what could have been a malicious attack, it would have been unwise of me to publish too much information, as it could have given the attacker clues and made the problem worse.

4. Availability rates in the 4 days up to the point at which the fix was implemented may well have been around the 86% to 90% mentioned.  During the period, we were doing (and publishing schedules for) intrusive testing which involved taking elements offline and then waiting to see if problems occurred.  You can't do intrusive testing without being intrusive!

5. All well run systems (especially servers) check their clocks against standard machines - "time servers" from time to time. I could 'hit' a machine every minute, within 1 or 2 seconds of a specified time in the minute on both machines.  Just like I can have the heartbeat kick in at the top of the minute, so could some other script kick ...

6. This is not a commercial site.  It is hosted on a server that has spare capacity and hosts some of my "day job" stuff.  The space and bandwidth and my time are provided free, and in such circumstances it is common practise to say "hosted by".   Actually, it's common practise to  have the "hosted by" rather more prominent, and to offer other adverts including graphics and logos ... placement 'above the fold'.  In other words, we (the coffeeshop) are well understating our free Web Space Provider.

7. Care has been taken, not only at the foot of every page but also duplicated in other appropriate places, to ensure that users are aware that they are not on a site run by or hosted by First Great Western. 

Tom's right - he IS being extremely picky and argumentative. But ... problem fixed, explanations and apologys offered, let's get on!

I see no further useful purpose in me going into the nitty gritty of the stats for the weekend when I was working on the server or earlier or how the issues were fixed.  Some of the other matters that are raised are worth coming back to - for example if members would like to see some sort of sponsorship, or adverts, or paid subscription or donation for a fuller service. But while we're just settling down after our most eventful week ever probably isn't the time.

Apart from Tom - who else wants to rake over the old coals?

[swlines]

Sorry, I'm known to be well known to not really be afraid of saying what I think of things.

I personally think that it is dodgy ground to be advertising the particular webpage that is on display - as it could be possibly interpreted that (and has been by others) it is there in order to promote 'making money' for a business other than FGW^▸ (First Great Western). While this can be seen as being good for the site (ie, paying for itself through this method) I feel it would be better done through the use of PayPal donations and Google Ads (I believe this was actually considered but rejected?). This would then cover the hosting costs of the site with ease I feel. I can't see this forum taking up very much space or bandwidth at all in absolute honesty (and remember, I've mentioned before that I've operated several forums so I know where I'm coming from).

The other issue I actually have is a conflict of interest issue but this is not the place to be discussing that (if Graham in particular (or any moderator) wishes to contact me on that subject to open discussions on that regard then they are more than welcome to do so and I will speak my concerns, however talking about the issue will drag the forum down further than it already has been in the past few weeks.).

Tom's right - he IS being extremely picky and argumentative.

Thank you for crediting me with this. I feel honoured. (er, if anyone has a sense of humour failure, that is a joke.).

[grahame]

Various models have been mentioned for covering costs, you are correct, and we do indeed wish to continue with one that not only has no conflicts of interest, but can also be seen by any doubting Thom(as)es to have no conflict.

Thank you, Tom.