Management Matt Posted March 10, 2022 Management Posted March 10, 2022 I'm currently working on your site (and was when you posted this 😅). I'll have it resolved in about an hour. Emediate 1
Management Matt Posted March 10, 2022 Management Posted March 10, 2022 I've resolved the issue with missing language strings. It looks like the language system has become confused and not all the language packs have the custom words. I'm not sure why this is. When a new language pack is added, any custom words are copied across. I wrote a custom script to fix this, and have run it on your site. The other issue with errors where stdClass::language() is logged is less clear. What is happening is that the member object isn't set up, which never happens, so I'm quite confused. I do notice that the donate app overloads the front dispatcher, so you may want to try disabling that for a while and seeing if the error returns. I also note that it's only logging issues from URLs sent via javascript and ajax. I also note that you get 4 or 5 errors then nothing for 15-20 mins. The errors only last for a single minute, so there may be a server process running? Something to investigate from your end. Emediate 1
Emediate Posted March 10, 2022 Author Posted March 10, 2022 Thank you Matt, sincerely. I am glad that the problem has been discovered in the core and it's obvious no third-party would have been able to discover this without your intimate knowledge of the software. I have taken your advice and disabled Donations, and will monitor the logs over 24 hours to see if there is progress there. I will pass on this information to our server techs too to investigate. Thanks again. Matt 1
Emediate Posted March 10, 2022 Author Posted March 10, 2022 (edited) Maybe we spoke too soon @Matt. Donations app has since been disabled, and all the same error logs (both types) have just returned. Edited March 10, 2022 by Emediate Matt 1
Management Matt Posted March 10, 2022 Management Posted March 10, 2022 Try disabling Feedback for a while too. Emediate 1
Emediate Posted March 10, 2022 Author Posted March 10, 2022 3 minutes ago, Matt said: Try disabling Feedback for a while too. Done (11:38PM 10/03 for my reference) Matt 1
Emediate Posted March 10, 2022 Author Posted March 10, 2022 @Matt still happening even with Feedback disabled.
Management Matt Posted March 10, 2022 Management Posted March 10, 2022 It's odd because the error indicates that code is running before the session has been read, which is impossible. It's worth asking your host to see if anything runs regularly that may trigger it.
Emediate Posted March 10, 2022 Author Posted March 10, 2022 I won't get that response until tomorrow (midnight here). OK to reenable those disabled apps in the meantime then?
Management Matt Posted March 10, 2022 Management Posted March 10, 2022 Yeah, sure - they don't look to be the problem. I'm looking at the code, and calling \IPS\Member::loggedIn() will set up sessions and set loggedIn() to a member via static::$loggedInMember = \IPS\Session::i()->member; so it doesn't make sense. It's not something we see in the software usually.
Emediate Posted March 11, 2022 Author Posted March 11, 2022 11 hours ago, Matt said: It's odd because the error indicates that code is running before the session has been read, which is impossible. It's worth asking your host to see if anything runs regularly that may trigger it. Hosts have checked server logs, and can't see anything out of the ordinary. We have a backup task that runs at 8:22PM and 2:00AM each day and lasts only a few minutes with only minimal impact on the server. Otherwise, nothing else of note.
Emediate Posted March 11, 2022 Author Posted March 11, 2022 On 3/9/2022 at 9:25 AM, Jim M said: If you duplicate (and give us access) to a test install without these third party applications/plugins, are you still receiving the same error? This would allow us to rule out the third party applications/plugins here. I think this is good common ground for us to move forward, don't you? @Matt - we are spinning up a test server and duplicated install to allow you to further investigate. Where can I send these access details?
Jim M Posted March 11, 2022 Posted March 11, 2022 18 hours ago, Emediate said: @Matt - we are spinning up a test server and duplicated install to allow you to further investigate. Where can I send these access details? You can place them in your Client Area access details area by visiting your client area, selecting the relevant purchase, then clicking "Review/Update Access Information" under the "Stored Access Information" section. We look forward to further assisting you.
Emediate Posted March 11, 2022 Author Posted March 11, 2022 It's not that simple unfortunately. This is a duplicated test server as requested - I don't want to change the stored access data for this site as there is another issue too that has been escalated to a ticket. The test server for this problem will require whoever is looking into it to modify their hosts file to access it.' It would be much easier if I could just continue this via email or PM directly with somebody please?
Jim M Posted March 11, 2022 Posted March 11, 2022 8 minutes ago, Emediate said: It's not that simple unfortunately. This is a duplicated test server as requested - I don't want to change the stored access data for this site as there is another issue too that has been escalated to a ticket. The test server for this problem will require whoever is looking into it to modify their hosts file to access it.' It would be much easier if I could just continue this via email or PM directly with somebody please? I’m afraid, we would not be able to modify our host file. Please temporarily assign a sub domain to it so we can access. You can use the Notes field in the access details area for adding credentials, etc… for the Test URL.
Emediate Posted March 11, 2022 Author Posted March 11, 2022 (edited) This really could not be any more difficult. It's costing me a small fortune already to have a duplicate server spun up, and all the data transferred. 😞 Edited March 11, 2022 by Emediate
Marc Posted March 14, 2022 Posted March 14, 2022 I can understand that, however we do need to be able to test fully without these items on there, so unfortunately is a necessary step in this case.
Management Matt Posted March 14, 2022 Management Posted March 14, 2022 Hi, I'm not sure what I can really investigate. It's not an issue with Invision Community and I'm not able to diagnose issues on your server unfortunately. All I can tell is that for some reason, sessions seem to be read later in the execution chain than they should. It's not something I've seen before with our platform. I've moved this into a ticket. Send in your test site details and we'll give it a once over just to be sure.
Emediate Posted March 14, 2022 Author Posted March 14, 2022 7 hours ago, Matt said: Hi, I'm not sure what I can really investigate. It's not an issue with Invision Community and I'm not able to diagnose issues on your server unfortunately. All I can tell is that for some reason, sessions seem to be read later in the execution chain than they should. It's not something I've seen before with our platform. I've moved this into a ticket. Send in your test site details and we'll give it a once over just to be sure. I was referring to @Jim M's comment that Support would not be able to modify their hosts file to be able to access our test server. We wanted to replicate the live install as much as possible. Configuring Invision to run on a different domain to that specified in the original installation is tricky at best. You may recall we had issues with this in the past and it was a nightmare. When bug finding, the last thing you want to do is introduce more errors and issues. I just found it annoying that Support would not be willing to make a 5 second edit to the hosts file in order to facilitate this investigation. In any case, the matter is redundant now, as you have stated your position as it not being an issue with Invision. I would have liked to know that before paying money and having the hosts spend a day spinning up a test server and duplicating our very large database over as requested in an earlier response. It is what it is. You can close this thread.
Management Matt Posted March 15, 2022 Management Posted March 15, 2022 Thanks for the update. I did mention that it was likely a server issue on a few occasions here: And here: Also, I believe you reached out to Stuart and he also confirmed it's likely a server session issue. I'm struggling to recall if we've come across this before, but it may be worth checking to see if your server runs Varnish and if so, if it uses any caching that can interfere with sessions.
Recommended Posts