Resolve "Many authentications requests coming from data2"

Closes #231 (closed)

@demariaa Sometimes the app enters a loop triggering many anonymous logins. At this time I have no idea what triggers this loop, where in the code it is, and how the app enters it, and cannot find any way to reproduce it.

However, this MR removes all the elements I can find that could potentially cause a loop in the authenticator system.

added 1 commit

• 5ea65915 - try some changes

Compare with previous version

marked this merge request as ready

changed the description

Would this explain the avalanche of authentications we have suffered or is it unrelated?

It could. I would need to analyse the logs from icatplus to make sure of it: You said there were 27k logins over the weekend from data2. The loop I observed from my network recording generated 1600 logins over 5 minutes. Where do the other ones come from? Are they other short 5-minutes spikes of login or are they spread over time? Do they come from many different IPs or not?

Then, it seems that there is a candidate!!

I did not see during the weekend such avalanche of logs, just many logs but pretty separated over the time.

If they are really spread over time, then there is another different problem. What I observered and am trying to fix here is a short avalanche of logins (1600 over 5 minutes in the case I recorded - But I do not know how and under what condition this issue starts and ends, so I assume it could also trigger shorter and longer avalanches).

It would be very useful to have a plot of the login times to see if they are forming groups of activity spikes (we should see at least one on the 1st of october at 11:26) or are really evenly spread over time.

Yes, but data2 seems to be the main suspect in both cases as the problem has not fully understand as you said.

It might happen to be the same problem but it did not happen during the WE or that I did not check them in deep enough (I did a grep and gave a quick look)

This issue is definitelly the main suspect for the avalanche problem but my point is, we need to check wether the 27k logins recorded over the weekend are the same problem or not. If they are many small avalanches, then we only have one problem, if they are evenly spread, then we have a second different problem.

Also, since I do not know how to reproduce the issue, I cannot test that the changes here solve it. But we can just deploy it and see if it still happens?

I've copied the logs on /tmp_14_days/demariaa

mentioned in merge request !212 (merged)

Replaced by !212 (merged)

closed