"Ongoing issues" rather than from last night's fiasco

... some of our pages / content include references to images held elsewhere, and many of them - especially old ones - are http rather than https.  Browsers barf at these in various ways, depending on which bowser it is and what security settings the user has made.    Although I have edited / changed to an alternative secure URL for images as I have come across them, that's an awful long process with a tiny return for each edit, and I have in mind a better solution as I do other things.

... there are occasional times where the worker has been too busy to accept connections from the receptionist - a new issue since December as we only a single server so no internal comms to go wrong before December. Our discussion on another thread about AI and crawler load rather points to the "culprit" issues and by telling or bashing certain of the crawlers we have an element of control.   Log files give me an insight as to the prooportion of theses we have;  as they "turn over" during the period the server was down, yesterday's and today's are obscured by those other things.  Come Monday morning I'll have further data.

As part of the analysis of last night, there is a possibility that I'll suggest / make a couple of server changes in the next couple of months to raise the roof and give us more headroom because of the 'bots.

... the "view unread" issue relates right back to the base SMF software we're running - always been a bit flakey and it's one of those things that effects some users irritatingly and others not at / not reproducible.   Blooming had things to debug, too, issues that report on last 24 hours.  Once I see the problem, it's gone away ... and another period to wait to see it again.     I suspect intermediate caching devices which double-dip sometimes and like my testing, the first time they get correct unseen flags.  Coming back a (milli)seconds later, there is then nothing new.  That's a theory that would explain what you see.  And our server does see double hits.

 
I'm still getting the unable to connect securely via https message occasionally on Edge and clicking on the view unread posts button often returns no items even though it's probably been 24 hours since I last viewed it.

 
Testing server further to make sure other bits are working ...

During the night, sitting up - "Dad - it's bed time"



and this morning ...

 


It would appear that no data / posts have been lost, and the server appears to be running normally at present.   I would recommend members to cut-and-paste and keep a local copy of anything they submit over coming hours while I come to understand what happened and take any steps needed to keep the Coffee Shop open as normal. 

I do NOT have any concerns with regards data security though (of course) I keep an eye open for any worries in that area.  I am concerned (until I know what went wrong) that it might happen again.  I await a ticket update from our web space provider; first concern had been to restore service.

 
One of the two servers that provide the Coffee Shop website and forum failed in some way at around 22:30 last night (29th August 2025).  I have been investigating and in touch with our web space (hosting) provider, and the server is up and running again (writing just after 06:00 on 30th August.

I am now into a learning and checking phase - I don't yet know what went wrong or whether the fix is a permanent one.  This post is something of a test in itself and I will update you all soon - need a cup of coffee first.