Archived Notification Feed

[ChloeTzang]

Aaand ... my Activity feed is now showing the "technical issue" error.

--Annie

Mine too and it's getting worse

I see a count if (4) under Works Activity but there's nothing actually showing and they're not showing in the archive either. I know one of them is a new comment on my latest story because it's there in Works, but I only know that because I went to the story.

In terms of a release, this is, bluntly, a fuck-up and should be rolled back. Applying fixes to a faulty production release is a panic move and generally only make things worse in my limited experience as a user. And I have done user acceptance testing at work on a couple of pilot rollouts.

Sorry Lit, but when it messes up like this, yu roll it back, figure out what's wrong, retest it and roll it out again later. You don't fix on the fly......

Anyhow, with the Pending Issue still there, I'm just going to hold off on submitting anything until it's fixed. I was lucky getting my last story go live eventually, I'm not going to push my luck.

 

[ChloeTzang]

I can only assume that the current problems include a failure to notify people when their favorite authors have published a story. The views and votes are extraordinary slow since my story was published.

That's going to impact the Halloween Competition too.

 

[TheWritingGroup]

I can only assume that the current problems include a failure to notify people when their favorite authors have published a story. The views and votes are extraordinary slow since my story was published.

You don't have to assume. The Activities screen is broken, so our followers are not being notified in a useful or reliable way. Thus, some of us (me) are not submitting new stories.

@Laurel is going to have a flood of submissions when they finally unbreak the notifications.

--Assume

 

[Jorunn]

Plenty of free time now to work on those 2026 competition and challenge stories. Nothing like leaves blowing in the wind, frosty mornings, and snow on the mountains to inspire a 2026 Summer Lovin story!

 

[EmilyMiller]

Sorry Lit, but when it messes up like this, yu roll it back, figure out what's wrong, retest it and roll it out again later.

You assume that they can roll it back of course.

 

[TheWritingGroup]

You assume that they can roll it back of course.

That is one horrible implication about competence there.

 

[onehitwanda]

1. The archiving of notifications speaks of putting a new database table (or more likely, set of tables) in play.
2. The shit performance since then speaks to me of someone not indexing the correct columns, or not understanding how to design a database at all.
3. They probably can't roll back because the table schemas are almost certainly different and they never thought they'd need to
4. There's no way to move history between the old and new structures and every minute that goes on causes a further divergence in worlds.

End result: The faberge egg they built and tested(*) on (probably**) at most a hundred test users with some small amount of activity between them is now committing loud, dramatic, tearful suicide front center stage while the audience undergo mass possession by Bile Demons.

Lit is not a big site - there are likely at most 100,000 records being created daily on the story side if you factor in voting, comments, stories etc. 100k. That's barely a busy five minutes in a modern trading system.

It's gone beyond embarrassing and well into farce.

(*) I'm charitably assuming some testing was done, though as the hours go by that assumption is aging like milk.
(**) assuming they tested with a copy of their production database, the current state of author side of things indicated that they did no automated load testing or concurrent user tests - an unforgivable oversight when dealing with live user traffic on a high-traffic site.

 

[ChloeTzang]

1. The archiving of notifications speaks of putting a new database table (or more likely, set of tables) in play.
2. The shit performance since then speaks to me of someone not indexing the correct columns, or not understanding how to design a database at all.
3. They probably can't roll back because the table schemas are almost certainly different and they never thought they'd need to
4. There's no way to move history between the old and new structures and every minute that goes on causes a further divergence in worlds.

End result: The faberge egg they built and tested(*) on (probably**) at most a hundred test users with some small amount of activity between them is now committing loud, dramatic, tearful suicide front center stage while the audience undergo mass possession by Bile Demons.

Ooooook. Yeah, so no parallel database loading going on. Just switch to the new and go for it. Yikes. No wonder we're not hearing back. And there is still the Pending stories black hole issue. And "AI" rejections.

Lit is not a big site - there are likely at most 100,000 records being created daily on the story side if you factor in voting, comments, stories etc. 100k. That's barely a busy five minutes in a modern trading system.

It's gone beyond embarrassing and well into farce.

(*) I'm charitably assuming some testing was done, though as the hours go by that assumption is aging like milk.
(**) assuming they tested with a copy of their production database, the current state of author side of things indicated that they did no automated load testing or concurrent user tests - an unforgivable oversight when dealing with live user traffic on a high-traffic site.

I getcha. I have IT friends, one of who works on SWIFT support for one of the big Canadian banks - she explained the volumes to me and I was, like, woooooooo. Compared to Health Benefit Admin Systems or Banks, the volume here is minor, but on the other hand, they have the teams to support those systems.

If Manu is being a jack of all trades and doing it all himself, I can see this happening.

I just wish they'd jump on here and ask for help. There are enough people here with the skills to actually jump in and look at things and fix 'em, I'm sure. Heck, I've done UAT testing and I'd volunteer to do UI testing on stuff like this for free. I like making developers cry. LOL. It's straightforward functional testing after all - nothing complex about it in terms of functionality. Back end and performance testing might be a bit different but I wouldn't have a clue how to do that anyhow LOL.

 

[EmilyMiller]

Lit is not a big site

I think people don’t realize this. But I think it’s demostrably true. I read all the time, ‘it’s such a massive site, cut them some slack.’ But it’s really not.

 

[onehitwanda]

Ooooook. Yeah, so no parallel database loading going on. Just switch to the new and go for it. Yikes. No wonder we're not hearing back. And there is still the Pending stories black hole issue. And "AI" rejections.



I getcha. I have IT friends, one of who works on SWIFT support for one of the big Canadian banks - she explained the volumes to me and I was, like, woooooooo. Compared to Health Benefit Admin Systems or Banks, the volume here is minor, but on the other hand, they have the teams to support those systems.

If Manu is being a jack of all trades and doing it all himself, I can see this happening.

I just wish they'd jump on here and ask for help. There are enough people here with the skills to actually jump in and look at things and fix 'em, I'm sure. Heck, I've done UAT testing and I'd volunteer to do UI testing on stuff like this for free. I like making developers cry. LOL. It's straightforward functional testing after all - nothing complex about it in terms of functionality. Back end and performance testing might be a bit different but I wouldn't have a clue how to do that anyhow LOL.

I'm busy building a site myself, in my spare time between herding cats at work and making you lot cry here. The very first thing I did was write up a document that lists use cases. The very next thing I did was create automated test cases that execute actual, LIVE requests against any code I write to verify that core systems are compliant. Create an object, verify that the state is as expected. Mutate it. Verify the changes have been persisted. Try to create it without mandatory parameters and verify that the calls failed.

The very next thing I will do when I've got the basic functionality implemented as a MVP is put together a suite of performance and load tests that I can run ALL THE TIME against the latest "release" of whatever I have.

This is not me trying to humblebrag that "I would do it better". This is a "These are the things that any competent full-stack engineer should simply do as a MATTER OF COURSE". Like memory items for a pilot, or knowledge of their ships draught and maneuvering speed for a Captain.

For shits and giggles I went looking for Lit's API backend.



This is a major red flag.

 

[ChloeTzang]

I'm busy building a site myself, in my spare time between herding cats at work and making you lot cry here. The very first thing I did was write up a document that lists use cases. The very next thing I did was create automated test cases that execute actual, LIVE requests against any code I write to verify that core systems are compliant. Create an object, verify that the state is as expected. Mutate it. Verify the changes have been persisted. Try to create it without mandatory parameters and verify that the calls failed.

The very next thing I will do when I've got the basic functionality implemented as a MVP is put together a suite of performance and load tests that I can run ALL THE TIME against the latest "release" of whatever I have.

I got pulled into UAT for an upgrade at work after it screwed up and got rolled back - it was more of an admin system rollout, not critical system stuff, and my boss asked me to do some quick checks before we started using it and i was in that WTF WTF WTF mode. LOL. So I got roped in to work with the dev/design/test team - that Agile Scrum stuff which was totally fun - I got into setting out business use cases - and I got to see what you mentioned above as well so I totally get that as well as the daily smoke tests and perf testing and stuff like that. It was really interesting but definitely a side-line. I know enough about it to know what you're talking about without having a clue about how to do it.

This is not me trying to humblebrag that "I would do it better". This is a "These are the things that any competent full-stack engineer should simply do as a MATTER OF COURSE". Like memory items for a pilot, or knowledge of their ships draught and maneuvering speed for a Captain.

Yup, gotcha

For shits and giggles I went looking for Lit's API backend.

This is a major red flag.

LOL I'm not even sure what the implications are of that. Blank look.

 

[onehitwanda]

I got pulled into UAT for an upgrade at work after it screwed up and got rolled back - it was more of an admin system rollout, not critical system stuff, and my boss asked me to do some quick checks before we started using it and i was in that WTF WTF WTF mode. LOL. So I got roped in to work with the dev/design/test team - that Agile Scrum stuff which was totally fun - I got into setting out business use cases - and I got to see what you mentioned above as well so I totally get that as well as the daily smoke tests and perf testing and stuff like that. It was really interesting but definitely a side-line. I know enough about it to know what you're talking about without having a clue about how to do it.


Yup, gotcha


LOL I'm not even sure what the implications are of that. Blank look.

The API that the java script (fuck off you censor cunt) website frontend queries to get the lists of things like activity, books, works, commets etc is timing out. That api call requests ten items from the activity list.

It's failing on the api server - which means that the api server is unable to cope with the load from the website. This means a couple of possible scenarios:

1. the database is poorly designed (<---- we are probably here)
2. the database is designed well, but is not tuned
3. the database is designed well, and tuned appropriately, but is still backing up due to a lack of processing power
4. the database is fine, but the api itself is badly coded.
4.1 it could be executing very badly designed SQL queries (<---- we could be here )
4.2 it could be doing everything in a single process, which means that all requests are FIFO and if you're at the back of the line, well, fuck you.
4.3 they could simply not have enough processing power available because their hosting provider doesn't support hot-swapping or in-place-upgrades.
4.4 they have no ability to add additional servers because lit's api cannot run as a high-availability system because it makes assumptions about how things work (<--- I hope to God we aren't here)
5. they're hitting fundamental limits with the underlying architecture of the programming language they're using.

 

[alohadave]

I can only assume that the current problems include a failure to notify people when their favorite authors have published a story. The views and votes are extraordinary slow since my story was published.

New stories not showing on author's page can't be helping. Most of the new stories aren't showing when they are published, so even if you are following an author, unless you have the direct link, you might not find it.

 

[EmilyMiller]

The API that the java script (fuck off you censor cunt) website frontend queries to get the lists of things like activity, books, works, commets etc is timing out. That api call requests ten items from the activity list.

It's failing on the api server - which means that the api server is unable to cope with the load from the website. This means a couple of possible scenarios:

1. the database is poorly designed (<---- we are probably here)
2. the database is designed well, but is not tuned
3. the database is designed well, and tuned appropriately, but is still backing up due to a lack of processing power
4. the database is fine, but the api itself is badly coded.
4.1 it could be executing very badly designed SQL queries (<---- we could be here )
4.2 it could be doing everything in a single process, which means that all requests are FIFO and if you're at the back of the line, well, fuck you.
4.3 they could simply not have enough processing power available because their hosting provider doesn't support hot-swapping or in-place-upgrades.
4.4 they have no ability to add additional servers because lit's api cannot run as a high-availability system because it makes assumptions about how things work (<--- I hope to God we aren't here)
5. they're hitting fundamental limits with the underlying architecture of the programming language they're using.

It’s kinda sad that I understood most of that. I’ve been the BPO for the transactional system my department uses for a while. A few months back - as a reward - I additionally got made BPO for a cross-org data repository. Seems like I’m doing more and more IT-adjacent work. And IT is just sticking to pure DevOps.

 

[iwatchus]

New stories not showing on author's page can't be helping. Most of the new stories aren't showing when they are published, so even if you are following an author, unless you have the direct link, you might not find it.

There has been a bug that new stories only show up in alpha order not date order. I look at date order 90+% of the time as would anybody looking for new stories.

It makes me queasy imagining how this is coded to end up with that bug

 

[Voyeurkenneth]

My control panel seems to be loading faster. My "All Activity" tab has a permanent parenthetical though, of the number of notifications I've had since the initial upgrade announcement. But the parenthetical won't go away on refresh. In other words, it's stuck on:

Same for anyone else?

 

[ShelbyDawn57]

The API that the java script (fuck off you censor cunt) website frontend queries to get the lists of things like activity, books, works, commets etc is timing out. That api call requests ten items from the activity list.

It's failing on the api server - which means that the api server is unable to cope with the load from the website. This means a couple of possible scenarios:

1. the database is poorly designed (<---- we are probably here)
2. the database is designed well, but is not tuned
3. the database is designed well, and tuned appropriately, but is still backing up due to a lack of processing power
4. the database is fine, but the api itself is badly coded.
4.1 it could be executing very badly designed SQL queries (<---- we could be here )
4.2 it could be doing everything in a single process, which means that all requests are FIFO and if you're at the back of the line, well, fuck you.
4.3 they could simply not have enough processing power available because their hosting provider doesn't support hot-swapping or in-place-upgrades.
4.4 they have no ability to add additional servers because lit's api cannot run as a high-availability system because it makes assumptions about how things work (<--- I hope to God we aren't here)
5. they're hitting fundamental limits with the underlying architecture of the programming language they're using.

OMG!!! Have I told you I love you lately???
Intelligence is so damned sexy, and that you can elucidate this so clearly, damn, you are the perfect woman.

I'm thinking that the simple fact we don't have one version of the truth(rating, view counts, etc., are different all over the site) tells me that 1 and 4 are definite candidates, I'd bet this weeks lunch money on 4.1, as well. None off what you mention is out of scope.

I honestly wish there were some way we could help.

 

[ShelbyDawn57]

And this from Manu this morning. Things seem to be resolving...

https://forum.literotica.com/thread...l-panel-activity-feeds.1642750/post-101709480

 

[TheLobster]

And this from Manu this morning. Things seem to be resolving...

https://forum.literotica.com/thread...l-panel-activity-feeds.1642750/post-101709480

I'm not sure about that. Right now my panel shows these tabs:



and clicking on either doesn't clear out the parenthesized count at all...

 

[NuclearFairy]

I haven't checked my feed so much since I first started visiting the forums semi regularly.

My feed seems to be working okay, without any of those annoying stuck parenthesizes. Have you guys tried refreshing the page? That's what did it for me.

 

[PennyThompson]

my activity feed doesn't seem to be stalling out anymore, but I haven't seen any notifications come through yet... somebody do something to me so I can see if it pings

 

[Bamagan]

my activity feed doesn't seem to be stalling out anymore, but I haven't seen any notifications come through yet... somebody do something to me so I can see if it pings

 

[PennyThompson]

View attachment 2571926

shh i'm building engagement

 

[PennyThompson]

thanks @Statius it worked!


Now we'll have to wait and see if the "my authors" feed starts working correctly...

 

[onehitwanda]

It makes me queasy imagining how this is coded to end up with that bug

If they've moved from an older model where they executed hand-written SQL against the db, to a model where they use an ORM (Object Relationship Manager) to query the database, it's entirely possible that the ORM orders via natural order for certain columns (like strings) - sensible in isolation, boneheaded for a site like Lit, and something that should be tested for regardless.