couldn't login, it wasn't just me!

[ChrisGreaves]

OK.
It is time for all of us to collect data.
Computers Process Data to get Information,
Good information can lead to Good Decisions, and Good Information depends on Good Data.

There is no doubt in my mind that the Powers That Be are aware of login problems.
I can see no point in reporting that "An hour ago I had a problem logging in".
But I believe that there a half dozen of us who would do our best to record "unable" dates/times in a format suitable to admins, and that the data might show a pattern against whatever other data admins have available to them.
A data pattern might show that Unables occur when
(1) logged in users exceeds some value
(2) specific times of day (UTC)
(3) specific users are logged in (maybe the lounge can detect when someone from Bonavista is logged in)
and so on.

A proposal: The administrative team let us plebs know exactly what data might be useful in looking for patterns, and some of us agree to log/record that data for upload in an attempt to determine where the problem appears to, er, be appearing.

I make this proposal confident in the belief that there are users here who know more than I about spreadsheet analysis.

Eileens Lounge is a machine that is a fixed mechanically defined procedure.
We just have to locate the problem.

Just like Alan M Turing (invented programming) and his bicycle chain.


Humbly
Chris

[viking33]

Uh huh. right.

[HansV]

I'm afraid that it is unpredictable. There have been times when I was locked out while others were posting without problems and vice versa....

[StuartR]

The admins have tried to find a pattern in the data we have, and it makes no sense. Sometimes I am locked out from multiple usernames, on multiple devices, from multiple networks - at the same time other admins are able to log in fine. I cannot imagine how the lounge knows not to let me in from any device with any username.

[Nick Vittum]

Issues with the server, I've assumed. I couldn't log in last night around 10 or 10:30 my time. That would be 2:00 or 2:30 GMT. UTC and GMT are the same, I think?

[HansV]

In winter, GMT = UTC. In summer, GMT = UTC+1

[StuartR]

In winter, GMT = UTC. In summer, GMT = UTC+1

Actually, GMT is always the same as UTC.
Here in the UK we use GMT during the winter, and BST (British Summer Time) in the summer

[HansV]

Thanks, I wasn't aware of that.

[BobH]

I just had a period of about 10 minutes where I couldn't log in.

I was logged in and reading posts then was logged out. When I tried to log in I could not. I tried several times pasting my credentials ergo no typos.

I got this screen saying that in order to log in I had to register.

login message.PNG

Would it be possible for this thread to be made a sticky so that folks could report the particulars of their login issues in hopes of providing a piece of data that might lead to a solution?

[HansV]

We (admins) have been observing this problem closely for over four months now, and we haven't been able to discover any regular pattern or common factor...

[Nick Vittum]

The times it's happened to me have always been after 10 PM, my time, so in the wee hours of the morning in Europe. Which is what led me to believe that that was the time when the server was down for, err, servicing.

[HansV]

No, the server is not down - while you were locked out, others were probably able to post without problems. I have definitely seen new posts appear while I was locked out, and it occurred at random times of the day. Sometimes it lasts only 15 minutes, but sometimes much longer...

[Nick Vittum]

No, the server is not down - while you were locked out, others were probably able to post without problems. I have definitely seen new posts appear while I was locked out, and it occurred at random times of the day. Sometimes it lasts only 15 minutes, but sometimes much longer...

Huh. Well, I'll just hope it keeps happening at night, for me. It's a good reminder to shut the computer down and go to bed.

[StuartR]

My longest lockout lasted about 26 hours! During that time other admins could log in, and post. I could not log in from any device or IP address

[Nick Vittum]

I haven't had much problem with getting locked out, yet, but I sure do get kicked out of the site and have to log in again. Is there any solution to that? (I'm in Firefox, latest edition, if that matters.)

[HansV]

Do you tick the check box "Remember me" or "Log me on automatically each visit" depending on where you log in from?
If yes, do you use an automatic cookie cleaner such as CCleaner?

[Nick Vittum]

Do you tick the check box "Remember me" or "Log me on automatically each visit" depending on where you log in from?
If yes, do you use an automatic cookie cleaner such as CCleaner?

Firefox cleans my cookies. I think I can set an exception for the site, if that will help. But I thought it only cleaned the cookies when I voluntarily log off. Seems like it shouldn't do that if I get kicked off, or timed out, or whatever's happening. It's not consistent. Sometimes I can have the site open but inactive for an hour or more, and I stay logged in. Other time, ten minutes of inactivity and I'm kicked off.

[ChrisGreaves]

I claim that right now (March 2020) the behaviour of the Login is unpredictable because we do not have enough Information. We obtain information, said Ian Sharpe, by Processing Data.

Treat The Lounge as a mechanical device. Unless there hardware or software faults triggered by a truly random event (cosmic ray, degrading atom of Uranium-2xy) then the event “cannot log in” must be caused by some external event. This may be as simple as a pattern of intervals (times between successive log ins) or it may be the hour-to-date sum of the number of characters typed in as usernames and passwords – a sort of Buffer Overrun bug, or whatever, but we must start by posing that there IS an event that triggers the bug.

By definition this event can not be triggered in a state of quiescence, just as a set or locked, parked cars in a parking lot can not trigger a collision; only when drivers start the engines are collisions a potential.
So the event must be triggered by something that Lounge Users do, starting (in any one person’s case) with the login itself, and ending (in that same person’s case) with the logout event (either initiated by the user, or by a system timeout after so-many-minutes of inactivity); this corresponds to the period “driver enters, then exist the car”. Any one or combination of user activities will be the cause of the “cannot log in” event. I note that this condition might arise some time after the user has logged out. This is somewhat similar to an outbreak of disease long after the initial carrier has departed the area.

So, here is my stand: The “cannot log in” event must be triggered by [can be traced back to] some user activity within the confines of The Lounge.

To locate that activity we need to search in the most likely area.
To narrow down the search area, we need to spot a pattern in earlier user behaviour.
That pattern can be determined only by obtaining system data for each failed (and I now believe each successful) login.

Successful log in data will be available from the system logs.
Failed login data will have to be collected by users, unless these failures too are recorded by the system.
I note that when I fail to log in, the login-screen is re-presented to me. If that refresh or repeat is recorded as the system asking for a log in, and if two such request do not result in a login, then that null login might serve to identify an event “cannot log in”
The unhappy user is not off the hook, for presumably the system does not know (a) length of username/password strings (b) current location of user (geographic site, whether home-based router or public WiFi, and so on).

We need to accept that there must be a well-defined mechanical cause of the problem, and then come up with a set of data items that might, when analysed, give us a clue to a pattern.

A pattern being identified, we might then refine our data definition to narrow down the search.

I have confidence in our administrators; I feel certain that they have discussed these “cannot log in” events, but without real data, how can the source be traced?

This is a call for our data analysts to suggest data that surrounds regular login/logout events, and then call for a commitment for those of us who care to maintain records of failed log in events and submit those records on a regular basis. I suppose that we need (groan) some sort of committee to come up with this plan, set a time limit, collect and analyse the data. Then there will be a refinement process in which the data is re-specified and we go through it again.

I do too suppose that administrators have communicated with the software supplier, and with its user group, if such a group exists, to see who else is suffering as we suffer.

Thanks
Chris

[ChrisGreaves]

The admins have tried to find a pattern in the data we have, and it makes no sense. .

Perhaps it is time to collect more detailed data from a broader market and give the patterns a chance to show up.
In short "A bigger/better sample of data"
Cheers
Chris

[ChrisGreaves]

Issues with the server, I've assumed. ...

I suspect so too, although right now I would say "issues with The Lounge".
Just not a random cosmic ray. Yet.
Cheers
Chris