Forum Outage

[Budgie]

There seems to have been a problem with one of the files that runs the forum this morning and it was showing a PHP parse error on when you came onto the forum home page.
I have aboslutly no idea what caused this as nothing changed in that file but I uploaded the backup copy and it's all working again.

So sorry about that.

[skyewright]

I have aboslutly no idea what caused this as nothing changed in that file but I uploaded the backup copy and it's all working again.

Funny, things computers...

Thnaks for getting it back up promptly. 

[Budgie]

Some of you may have noticed that the forum was down this morning (Friday).
This was due to a server problem at the host. They had memory issues with the server we were on and had to swap us onto a new (updated) server but the process created an error on their MySQL database which we use to run the forum. This have now been fixed and I hope it didn't cause any issues with an of you. 

I have support ticket logged with them at the moment as my clientraw.txt files are not being renamed as they should be using the normal WDL FTP programme. I'm having to upload on the normal WD FTP programme instead and that's timed at 5 minute intervals so the 'live' data on my site isn't working just now. 
I just turned off the renaming of the file and it's uploading normally now, but I still want them to fix the issue.

[munrobaggins]

Some of you may have noticed that the forum was down this morning (Friday).

I was wondering what had happened...all's well that ends well...as they say 

[skyewright]

I just turned off the renaming of the file and it's uploading normally now, but I still want them to fix the issue.

You may well be aware but just in case...

At 0912 BST you are shown on the map as "Offline" both on my own SCWN page and and on the map page on the SCWN site.

[Budgie]

I think that everyone is offline at the moment.
I just looked at the cache file and the times for the last update of all the clientraw files was 11:02 on the 18th.
If you check your data on SCWN to your current data then you'll see it doesn't match. 

I'll see what I can do.

[Budgie]

Right, after further investigation it looks like the cron job that tells the server when to collect the data from everyone's websites has stopped working.
I have added this to the support ticket I raised yesterday with the host. It maybe that the cron tabs on this new server require a slightly different coding but I'll have to wait for them to get back to me before I'll know anything.

[skyewright]

I think that everyone is offline at the moment.

Ah. I see. Most are showing as "present" on my view, but are certainly not "current".

Conditions data shown was collected
from Fri, 18-Sep-2009 17:00:43 BST to Fri, 18-Sep-2009 17:17:13 BST

Best of luck with getting a fix...

[Budgie]

I got a reply to my support ticket and they forgot to start the crontab service after the move, he said that he would restart it at 11am this morning but still nothing yet.

[Budgie]

Looking into the script a bit further, I think this is more to do with the renaming issue that I'm having then it is the crontabs.
I've just checked the process for getting the data and it requires the SCWN-cache file to be assigned as a .tmp file and then it's converted. This is the same thing that my own clientraw file goes through, so if there's a rename issue on the new server then the SCWN-cache file isn't getting renamed either and hence we're not getting the updates.

Back on the emails to UKServers again. 

[carseman]

It all seems to be working again @ 21:00hrs.

Chris

[skyewright]

It all seems to be working again @ 21:00hrs.

Here too @ 2110.

I noticed that earlier on Martin had added a temporary "there's a problem" banner to the SCWN mesomap page.

I wonder if the php behind the standard mesomap page could gain a bit of extra "intelligence" to include something of the sort automatically if the data had become obviously "stale", e.g. if the "to" time in the "Conditions data shown was collected" bit is more than some threshold, say an hour?

The time range is there to see, but if people are used to data being current, they may well not look at the "Conditions data shown was collected" for some time, whereas a banner that's not usually there would stand out.

Just a thought. Hopefully this sort of problem is going to be as rare in the future as it has been in the past. 

[Budgie]

Yes, we seem to be back to normal again and the weather data & charts are now working again.
The issue was the crontab in the end.
First they forgot to start it and in the mean time I had uploaded a backup cron file that had a carriage return in it, so when the cron was restarted it still didn't work because of the file.
Still we're back and I hope it doesn't happen again!!

As for the renaming issue with my clientraw files, the new server won't overwrite and old file so when renaming, it wants the old one deleted first, so I've had to disable the renaming in the WDL FTP and will have to live with that for now.