I'm pretty sure you're aware of this issue, but nevertheless here goes:

Sometimes capital ships (in my case it was a trident) get "stuck". I'm not sure how the trident in my escort got stuck in Verasi O7 (as I left the sector after the ships entered), but all the other ships in the escort managed to dock in the sector, while the one cappy never left the group, keeping the mission open for over 45 minutes before I finally logged.

This incident happened earlier today. I don't remember the exact time, but it was the last mission I took today.

Perhaps the sector shutdown - timed out after you left.

It might fix things if mission npcs were considered "active" and prevented a sector from shutting down untill they left.

The cheap fix is to keep all sectors active for 10 minutes instead of the current 5 but that would increase the server load a bit.

Another odd thing I've noticed during an escort mission is that instead of docking, several of the NPC's will behave as if logging off in space. Ie, they hold stock still for 10 seconds and disappear without a trace. The mission still completed after this, though.

My understanding is that even when a sector is shut down, NPC's will carry on *performing* in an instanced (but offlne, none the less) sector. I may be wrong about this, as the entire system was re-written cos Debilitator was borked, and Kourier was about as reliable as UPS.

IE, sectors timing out should not affect NPC's. They should be allowed to finish what they are doing either way. CTC bots are a good example, they often travel through inactive sectors, but still reach their goal (or become un-instanced if an attacker shows up), so if an NPC is stuck, something deeper is wrong than a sector time out.

[edit]

Ahh, I should have said Debilitator was about as reliable as UPS for delivery, so they got a new Kourier, ;)

I know they are not suppose to be affected but evidence would suggest they are in fact being lost when the sector shuts down.

To be clear, I have no hard evidence that the sector shut down. However, that seems a possible cause given the outcome. Then again, black box debugging can only get you so far...

True, I can only guess at the problem.... But I cant help it - Its how I make a living.

Running approximately 50x as many sector daemons is not a practical option, and this bug has nothing to do with the sector shutting down anyway. I wasn't able to actually reproduce it in my testing, but there was an obviously-possible race-condition, which would result in this behavior; it's fixed for Friday.

So you may have fixed a possible bug?

I always report those bug solutions as PFM -Pure Freakin Magic.

At least your guess is a better informed one... Hope it works.

I made what was definitely unreliable code, reliable, and it is the only bit of code I can imagine causing the reported problem.

Since the bug is a race-condition, which requires other parts of the code to be a bit slower than they ever are on our development hardware, it's not surprising that I didn't see it happen after only about 100 automated missions. I could force the race condition that I think is happening to happen, but that wouldn't really prove anything. The only sensible thing is to release my fix and wait for you guys to tell me if the issue in question has gone away.

I was just pulling your chain....

Seems to be happening again. Though this time I'm not sure it's a cappy that's stuck. The last remaining ship in the escort has been hanging in Azek D11 for the last 10 min or so. I'll go poke the sector to see what's up.

Okay, it was a moth I think that got stuck this time. Within 15 seconds of entering the sector the escort completed (about 20-30 minutes after the other ships had all docked). I guess the sector shut down before mission complete?

It's happening again (right now). This time the entire escort seems to be stuck at the destination (Dau D-9).

again, the problem was fixed by someone entering the destination sector. within a minute of the sector being kick-started every escort had docked and the mission ended with success.

So many chains all dangling like little pieces of yarn.....

Im my profession its called voodoo maintenance. Its where you apply a fix to a problem without really knowing whats wrong or how the fix even works. While the "fix" works most of the time, it becomes a habit and over time it actually starts to take up more time then if we had done a proper diagnostic....I keep a rubber chicken on my tool box to remind myself of the danger inherent to the approach.

Intermittant bugs are the hardest to catch. Perhaps it might be possible to slip a few lines of code in to alert you when a voy takes to long and have it dump the sector state for you....

I sincerely do hope you find this as it has been a problem in one form or anouther as long as I have been here....

Thanks for the further reporting, Aticephyr.

PaKettle: I don't know that that the capship-sitting-at-the-end-bug was absolutely caused by the race-condition I fixed, but then, even if I had been able to reproduce it, I still wouldn't absolutely know that it was the only cause. I do understand the race-condition and my fix for it. Most developers would just say "fixed" in this situation; please don't make me regret my openness.

And please don't try to extrapolate the state of our codebase from symptoms, or lump subtly different symptoms together and think it's one bug, that we just haven't gotten around to fixing, and have had plenty of reports for; it's really counter-productive. One would think that other developers or systems-thinkers would be the best bug-reporters, but when you report things (or not) based on false assumptions or inferences, you just confuse the matter. This thread includes discussion of no less than four completely different bugs, and the only one that has existed for a "as long as [you've] been here" is with the non-trivial problem of pathfinding.

Sorry mom :(. If it happens again should I make a new thread and just link against this one?

edit: Just saw your edit. I'll keep reporting here if I see it.

A new thread would be better actually. I think of this one as the "capships' fake docking unreliability" thread. So, reuse it if you want, if you see that specific situation: capship in the destination sector, not hung up on a roid, last in the convoy, mission not ending. When in doubt, make a new one.

Unfortunately I wasn't able to find the cause of the one you reported most recently, from the logs, but I'm thinking about ways to log what I may need. It's always helpful to have a live example, so please report it immediately, if you run into one of these problems, and try not to disrupt it until I have a chance to look.

Uh actually you have had the premature sector shut down bug for at least a year. Previously it wasnt annoying enough to be worth my remembering since it usually only resulted in lost ore drops for the hive...

I wasnt trying to be critical and I do apologise if I sounded that way. It is rather easy to get hung up on a symptom and sometimes it helps to reasses the situation for clues that may have been overlooked.

All bug reports are full of bias and opinion...Its just the nature of the beast. I can only tell you how the code appears to be working and wish you happy hunting.

BTW - the original problem was the cappy never left the first sector and got left behind.....It is not the dock pathfinding problem...