I'm following with interest.

One thing to look out for is any timeouts. Latency on the trunk networks is now getting chronic, particularly at times of day when people are doing extensive video downloads. Contention ratios get overwhelmed and a server might time out after waiting what it considers to be too long for some acknowledgment from a user. I believe some ISPs are giving priority to video (as a sales leader) so non-video traffic carries the burden of erratic latency.

I've been having problems with major sites such as Google and Yahoo timing out on me, though my local service is supposed to be 8/1Mb/s. ISPs are promising more than they can deliver with any consistency, almost entirely due to the lumpy nature of video download traffic.

The web-server company that hosts 3 sites I manage went down for 16 hours last weekend after an equipment "upgrade" went wrong.

You are not alone!

Maven

I did note a slowness with the site late last night but that happens at times so I just put it down to here and the net.. rather then there...page took forever to come up but did fairly quickly loading time wise when they did load...long pauses I guess would explain that better...like 20~ 30 sec or more

I guess its as they say look at what you changed as probably being the cause (or something with in most likely) if it worked OK before...

when I logged off thought it was still up...

Bummer mate hate to see you go through these teething problems with the new version ...

if I notice it dies I'll send an email regardless I Guess the sooner one arrives the sooner you can get on to it and I tend to be on at odd times

PS your probably way already on to this but I'll say it anyway
since you say you do get a lot of hits from those that need castrating maybe see if your logs show any commonalities in the times the site dies

Maybe the new scripts have a floor that wasn't there in your old ones or even innocent hits maybe clobbering something but me no idea

Hope for you its a quick fix mate

I think I have sorted things out. It appeared to be a coding mistake I made when redirecting all traffic to the HTTPS protocol (this is a secure site now).

I won't be able to say for sure until I can see the site not crashing again. If I'm right I'll explain the issue in full.

In the meantime, business as per usual.

So far, so good. In the last few days this site has been getting hammered by the robots from seven search engines and whilst the timeout issue wasn't directly caused by them (the server can handle 50 times the punishment normally) it is looking like a mistake I made in a recent upgrade to the site caused the robots to overload the code snippet I modified.

Anyway, I won't say too much before knowing whether my fix has done the trick.

Yeah I was thinking along those lines..."hammered" actually Id kinda forgotten you did this now as secure site...which is good!...

Yeah coding not that I do much or know much these days.. but it can be real fun...(meaning nightmare)
with bugs that find there way in..fix or find a better way of doing something but break something else
in the process

on search engines thought it does dump me back to the main page now (or was over last few days) ...thought I haven't really thought about it....just made a mental note

when searching this site pops up a lot with specific Oz radios...thats a good thing I suppose if you want this site out there..but will probably annoy people if it no longer lands on the page they were searching... for information on

I probably should have mention it now coming to think of it and maybe that will just be a time thing with search engines re-finding the new links over time

Anyway fingers crossed it fixed


Cheers mate

As the saying goes in maintenance circles: when something otherwise reliable suddenly stops running, look at the last thing done to it.

Yes GTC so true and never change 2 things at once if you trying to figure out whats gone wrong

Normally that works though I am working on three upgrades (upload photos, thread move feature and another administration feature) at the one time at the moment and it's hard to keep track of things.

Two things are in my favour though:-

1. I religiously back things up before changing anything.
2. I make sure this site logs everything that happens and that did assist where the web server logs failed. The server logs only show "unspecified error" where things stopped working. The problem with this is that when a HTTP 500 Error shows on a visitor's browser, it can be due one of about fifty different reasons. Assuming one can discover the reason for the failure, there is still the task of locating the smallest mistake in a good 50,000 lines of VBScript and VB.NET code. The site's own logging system lead me to what hopefully was the stuffup though and as it turns out, it was only a recent change. The irony is that without the attention the site has been getting from the search engines this week I may never have discovered what I believe the problem was.

With regard to the search engines, I think they are all re-indexing the whole site because of the changeover to SSL and the HTTPS protocol.

I am working on three upgrades (upload photos, ...




... and another administration feature)

While you're at it, might I suggest a Contact Administrator entry in one of the strip menus. It's one of those things that newbies keep asking about how to do.

The search engines seem to be doing the job just checked a few searches..ALL GOOD

Yeah I notice a post or two on that GTC and agree its probably a good idea but I guess the reason why its not is bots if its in public view..thought Brad can answer that one for sure

Maybe a "contact admin" button at the top of control panel and profile with a bold note in the public view on were it is

or possibly if thats not public...under the recent forum activity part of the page

Anyway just a thought

I think the problem even thought its there plain as your face in the "quick reply" as you said in a thread, people just get use to expecting things to be in certain places..its probably really the reason win XP wont die or 7 people are just so use to its style and know were things are 8 just confuses them

Thats kinda what I meant by innocent hits, thought I didnt think of search engines but yeah I suppose there were plenty...glad it helped (hopefully) nail the problem thought..one good use for them for an admin

PS yes back up back up back up, the day you dont is the day you'll need it

...the day you dont is the day you'll need it

Been there twice, though with the backups. In the approx. eleven years the site has operated we've had two total server failures. Backups are the only way back.

Yeah I could relate a lovely story of two days worth of work having to be re entered cause someone decided to do something and for what every reason decided not to back up before they did (it wasn't me we were just the ones that had to help fix the issue)

I am one who should head his own advice a little better then I do but I do tend to try and have at 3 copies of things stashed on different drives like photos etc

I do Image my working drives its the quickest way to get back up if something fails

Anyway Brad glad to here you tread it like a religion cause really I guess you need to be like that with servers

Cheers mate

It'd be best if I was still using the servers pictured on the 'About' page. They are bulletproof and have two of everything. Two discs, two power supplies, two network cards, etc. Everything also works faster.

Unfortunately they just chew too much electricity in today's high price environment. One of discs on one of those servers threw a shoe once and it was just a matter of sliding it out and sliding in a spare. The working disc then copies itself to the new one via the RAID controller. All that equipment is in mothballs at the moment, in the vain hope that one day electricity will be inexpensive again.

Okay, I think it may be safe to say the worst is behind us. The site has been stable for more than 24 hours and it looks like it may remain that way from here on.

The correction I made yesterday was to prevent the search engine robots being locked in an infinite loop after arriving at the HTTP 404 Error page. Normally, this page should display the 'page not found' message for ten seconds then redirect the visitor (including robots) back to the front page.

Instead the 404 Error page was redirecting the visitor back to that page. With a few hundred instances of robots from Bing, Google, Yahoo, Yandex, Baidu, Majestic and a few other fairly insignificant search engines getting trapped in this mire, it caused the site to time out.

I discovered this when I noticed Google's robot looking for a file called 'robots.txt', a file which I do not use here since most search engines don't recognise it or obey its instructions. Anyway, patching the redirect code and rebooting the server to cancel out all the infinite looping seems to have fixed the problem.

Let's hope so anyway. I will be away for a few days and expect the site to be up when I return.