Science & Industry

So I'm manufacturing Gel-Matrix Biopaste, takes 3 P2 products to make. I've got my factories set up in hexagons, with the factories making the P2 dumping it directly into the P3 factory in the center, and the P3 factory dumping it back into the Launchpad.

The issue is, for some weird reason, one of my P2 factories isn't getting it's resources in a timely manner. I've triple and quadruple checked the routes (and reset them several times), and the materials are being routed there. However, if I leave it overnight, that one factory almost invariably ends up between 20 and 40 minutes behind all of the rest, which delays that P3 factory as well.

What on earth could be causing this? A screencap of the configuration is included below. All of the P1 is put in the launchpad, routed directly from there to the P2 factories (which transits through the central P3 factory in each ring, plus the P2 factory between the P3 and the launchpad), and the P2 is routed directly back to the P3 factory in the center. It's only ever that one factory, all the rest work perfectly.

And no, I'm not starving them for materials, either. I put all of my materials in the launchpad in 240 unit increments (the amount of each P1 my 18 P1 factories consume per hour), and I keep it topped off. The Launchpad has room for precisely 18 hours of production (4320 units each of 6 P1 products), and I refill it roughly every 12. It's currently empty because the only way I've found of "resetting" the P3 production is to starve the entire complex, then put the P1 back in the launchpad so they sync up again.

Screenie: http://img15.imageshack.us/img15/4672/bquh.png

Paste is produced at a rate of 3 per factory per hour. This char has produced 561 over the last 64 hours, while another of my characters has produced 576. The 15 unit discrepancy might be a result of me picking up the P3 on different cycles (which would explain 9 of the 15), and the delay introduced as I've reset the sync on the factories twice since I last ran materials out to those POCOs. The pair of those effects could reproduce a 15 unit loss over that timeframe.

Even if it didn't, a 15 unit loss is 7.8%, which would imply a 5 minute "hold" per cycle, on average. I've only ever observed it at a 20 or 40 minute delay off the other factories. In addition, it's consuming P1 at the normal rate, so it's never actually getting "lapped" by the other factories (never more than a single partial cycle behind).

Lastly, I've personally observed that single factory at the end of a cycle. It's output is properly being received by the P3 factory, and the new materials are properly being input instantaneously. The fact that the P3 factory is perfectly synced with this one delayed P2 factory suggests that that factory is in fact properly routing as well, so I'm honestly not sure where the disconnect is.



I've not tried swapping it to another P2 to test. Another of my chars (also producing Biopaste) had this problem a couple weeks back, though, on a different type of P2 factory. In that case, resetting the route worked, and I found out in the process that that route was weirdly set to route through another one of the radial P2 factories before going to the delayed one (so launchpad -> P2 factory link -> P3 factory -> another P2 factory -> P3 factory -> delayed P2 factory destination). Resetting that route seems to have fixed the issue on that char. However, no similar route is present on this character, and I've reset the routes to and from that factory several times.



I'm not sure if it was the last built, that was a couple months ago. It was not the last one I linked, nor the last one I routed (until I started messing with resetting the routes on it).



When under observation, it seems to work perfectly as expected. It's only when I leave it overnight that I tend to see these type of delays.

That is really bizarre behavior. Direct routing has been historically problematic, such that most people don't recommend doing it. However, any 'delay' caused by problems with that would result from materials being lost in the transfer process. From what you've said, that doesn't sound like it's happening.

If it were me, here are the things I'd try:

Delete and rebuild the problem factory, as well as one other factory.

Build the problem factory first such that is not the 'last' one built, and position the other factory such that it is both the newest and the furthest from the P3 in terms of links.

Swap the P2 production of these two rebuilt factories.

Run another, say, 6-12 hr cycle and see if anything is different.

If that didn't work, I'd try a different setup. Something like this, where there is no direct routing and everything goes through the Launchpad. It's not as pretty but you have the grid for it.

Completely up to you whether to try that. The behavior doesn't make sense to me. It feels like it has to be a routing/link problem or a bug with the factory itself.

I noticed a similar bug this weekend piping P2 into my P3 factories. I had 3 Guidance System factories and 6 Transmitter + 6 Water-cooled CPU factories feeding it. My 3rd P3 factory was always short 10 transmitters, as if i was only running 4 P2 factories instead of 6.

I created a 7th Transmitter factory after triple checking all my routes and what do you know, i now have 5 extra transmitters per cycle and the 3rd factory is producing like it should. I'll try removing the 7th factory and see if it bugs out again.

That actually make a lot of sense. It's absolutely horrible design (particularly if the polling interval is that long), but it does in fact make sense.

That said, this occurs even when the destination factory starts completely empty of materials, which weakens that conception a bit.

Seems to be working properly so far. I'm not sure if that's because I reset the routes, or because I babysat it through it's first cycle ending, but one of the two fixed the issue, or at least seems to have thus far.

That's why factories have input buffers. I'm not pre-loading that P3 factory, so even if the P1->P2 factories finish first, it just goes into the input buffer on the P3 factory. Since P2 and P3 factories are the same type, they have the same cycle length, and thus it's impossible for the P2 factories to cycle twice before the P3 cycles. Also, I don't have any basic factories on that planet. All of my P1 is manufactured on my ECU planets, not my factory planet.

In addition, as I've stated before in this thread, I've confirmed no materials are being destroyed. It would be very obvious if this were the case, because that would mean the input buffers on my P3 factories would be haphazardly filled, rather than completely empty. Excepting when that one factory was getting out of sync, their input buffers are always empty, so nothing is ever getting destroyed.

Lastly, as I stated above, routing through the launchpad is inefficient, because routing logic is bonkers. It prioritizes filling the input buffer on an already cycling factory over feeding needed supplies to a factory that's paused due to lack of said supplies. As a result, I lose total production output if I do it that way.



Precisely why I direct-route. Since my P3 factories are starting completely empty, their input buffers are also empty, so it will never be the case that any of my P2 gets wasted, because it will always be going into an empty input buffer.

Just checked, and it is.

So basically, the conclusion seems to be that there's just some weird issue with direct-routing and completing production jobs unless the factory is baby-sat. /shrug.