| Author |
Thread Statistics | Show CCP posts - 1 post(s) |
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
   |
Posted - 2010.06.11 18:25:00 -
[1]
I need to do more testing but I'm definitely getting material loss when I'm doing controlled production. At this point I suspect that it is caused by downtime causing the loss of the output side of a processor cycle while the input side keeps working. So for example when I start with 800 units each of an P1 material (going from storage to facility to storage) and I use a schematic to transform it into a P2 at an advanced production facility I end up with 95 units of the P2. Based on what I saw last night as I was starting the job and what I saw this morning after downtime but before the job finished, I'm suspecting that downtime is the culprit. I'll be doing some tests to confirm things over the next few days.
I also need to test this on basic factories. I dread having to test this on the high tech factories (gah waste of good P3 stuff that way).
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.12 04:19:00 -
[2]
experiment 1: Put one hour of reactants in a storage bin. that has no outgoing routes. Set the outgoing routes to an advanced factory. hit submit and come back in an hour
Results: expected results (i.e. 40 of each P1 -> 5 P2
Experiment 2: Put two hours - same as above
Results: expected results ( 80 -> 10 )
Experiment 3: Put 3h - same as before.
Results: pending
(note that I've now had 4 jobs that were 800 of each -> 95 and all 4 jobs started before and finished after downtime.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.12 10:19:00 -
[3]
Well, I never installed a client for Sisi. Meanwhile on with the experiment.
result 3: expected: 120 (x2) P1 -> 15 P2
Experiment 4: 1 cycle starting 55min before downtime and a separate one 2 cycles also starting 5 min before downtime. Both these should have a delivery during downtime. This is the litmus test to confirm the loss is occurring due to non-delivery during downtime.
result 4: pending.
As before the method to ensure start time is simple: I have 2 advanced production facilities connected to a spaceport. I clear all outgoing routes but leave the incoming routes alone at the spaceport. Then I import/expedite transfer required materials for the number of cycles of each input material for the two schematics I have setup in the two processors. Then I establish all 4 outbound routes at the spaceport waiting until the correct time to hit submit. This allows for great control over the sequencing and timing of the production.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.13 16:29:00 -
[4]
Ok I wasn't able to reproduce the problem with the short runs in experiment 4. However I was able to reproduce the problem with some advanced industrial facilities who's cycles were out of sync. The results on the other side of downtime showed the cycles now sync'ed and some input materials missing. Determined by the output count - i.e. the output was what I would expect by that point in time but the input side was suddenly short on 2 of the schematics. I assume the materials got blown up either in the current production cycle or on the input hoppers. Probably it killed the existing production cycle on those that it was syncing with.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.14 17:12:00 -
[5]
 Originally by: CCP Zymurgist
Hi Letrange, I hope you have submitted your Bug Report on this 
Indeed I have (#96977) although I'm going to have to continue testing as it seems not to be related to downtime, but to your cycle synchronization method. Whatever is doing the cycle synchronization is causing the material loss. I just had a loss that started and finished outside of downtime so I need to change that part. It's definitely related to the synchronization though. Grrr more testing to get a 100% reproducable bug... I'm on track though.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.16 19:44:00 -
[6]
well last night I did some short runs but I did them after I had all the extractors already running and with incoming routes coming from different storage than the outgoing routes were desined to. No material loss under these conditions. However I'll need to test doing the extractor scanning during and after assigning production routes and also to source and destination to from/to the same storage (both conditions I think were operating when I had the material loss).
Sorry it's taking so long to nail down the issue but kinda tricky to test with the delays setup in the system.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.17 15:33:00 -
[7]
Frustrating. I'm still trying to reproduce the conditions but I keep failing. There keeps being un-explained losses in the system. BTW it happens at all levels of production. It probably happens at the raw->p1 level as well but since the quantities are so random for that we usually don't notice it anyways.
It seems to be related to whatever mechanism is resetting the extractors from time to time.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.21 13:52:00 -
[8]
At least I'm not alone. I just had a sequence of events last night that resulted in both a re-set of the extractors and some material loss last night (ugh 6 P3 out the window). None of my other planets get re-set. Just this one. Also I had setup some schematics and their outbound routes without setting up some inbound routs yet (like to set that up just when I move the materials to the "ongoing proc storage").
The interesting thing is that the night before I noticed the loss I wanted to check some link volumes. So I setup a route in edit mode from the space port. But I never hit submit. I hit cancel. Now I thought I verified the materials and they were not gone. But this morning I shipped some totally un-related product up to the customs office and when I checked the quantities for production suddenly there was a cycle of materials missing. of exactly the item I used for that route test! And the extractors got re-set and the re-set time seems to correspond to the moment I hit the cancel. Of course I never bothered to verify the extractors after I hit cancel on the route setup.
At least this gives more more clues as to how to reproduce this bug. Still not reliably reproduced but I feel like I'm narrowing down on the problem.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.22 14:38:00 -
[9]
Welp - another good causality theory shot down in flames.
On the other hand I just got a rather interesting screen shot....
Blog Entry with pics
So as you can see there's a point where the inventory on screen shows more than the server thinks are there. The question is: without an outbound Route - where did those 20 units go? Um... CCP... Just how corrupt is that database? cuz atm a modi****of data consitency would be nice....
Also thanks for resetting my extractors (3 days in a row I haven't needed to restart my 24h extractors...). I must admit that's one bug I'm not complaining too much about although it does throw off my schedule a bit...
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.23 03:13:00 -
[10]
Ooooo Looks like we may have multiple causes of material loss. One of them definitively is related to downtime. See the following blog entry (with pics for proof)
Blog entry with MORE pics of bug
Note that during a single downtime - because of the production geometry - I lost TWO! full cycles during this downtime. Also note the time stamps I overlay-ed into the lower left hand corner of the crops.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.25 15:23:00 -
[11]
Well, I only lost a little bit of material to the seriously extended downtime. Looks like it was just the last cycle on the P3. So I was able to do a single P3 cycle to finish off.
Then I set up the P4 run (last P3 feeding material to the P4 cycle). Ouchy. Just checked this morning and even though I tried to time it to straddle the downtime, downtime must have been full sized today since I got nailed for 2 P4 items. Thanks CCP. I notice my bug report is still not filtered. {sarcasm}Glad to see the bug hunting is going well on the new content...{/sarcasm}
Side comment: I don't know if Sisi's downtime is identical to TQ but this may not be reproducible on Sisi, which would make the whole bug hunting thing useless in this case. The lack of response other than "submit a bug report" on this leads me to believe it may be a while before we any progress on this if ever. Brilliant. If I sound ****ed it's because the method of getting around the bug involves waking up early to start stuff right after downtime... Fraking inconvenient...
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.25 17:48:00 -
[12]
Originally by: Reygrimm
Originally by: Qwert0
I just did a small scale test of P4 production to get an idea of the set up for it. Stuck 60 of each P3 into the spaceport, and when i came back a few hours later I got only 5 of the p4 back, instead of the 6 I should have.
Qwert0, You should have had 10 I think? The P4s need 6 of each P3 item. 60/6=10. Unless there's something I'm missing here...
That's actually kinda interesting if you got back exactly half (5) of what you should have gotten (10). If you could reproduce it, might go a ways to get a simple scenario to help track down the bug.
Keep the faith guys!
You are correct sir, he should have had 10 at the end. Most of the losses happen during downtime where various strange and unsavory things seem to be happening.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.27 13:34:00 -
[13]
Edited by: Letrange on 27/06/2010 13:34:08
Originally by: Narfas Deteis
Confirming that I loose 1 output from each factory, so 1 unit of P4 disappears from each installation and then P4 manufacturing plant has to wait for another cycle of P3 manuf. plants and sometimes even P3 manufacturing plants have to wait for another cycle of P2 production.
Result: 2-3 units of P4 lost after each downtime for each P4 installation. It's around 20 P4s in my case.
No information concerning this bug in patch notes on Sisi.
If you or someone who's got sisi configured could see if this is reproducible on sisi and bug report it if it is, that would be nice. Not sure sisi's downtime is the same as TQ's which may be part of the problem. The bug hunter was asking if this was already fixed - I indicated that still not as of two days ago. I suspect they'll miss this one this patch round.
|
Letrange
Minmatar
Chaosstorm Corporation
Apoapsis Multiversal Consortium
|
Posted - 2010.06.28 14:17:00 -
[14]
Originally by: Qwert0
You are right, for some reason I had it in my head that it took 10 units, not 6. One thing to note is that this did NOT occur over a downtime, but just randomly while I was afk.
I have seen that. Mind you it's much rarer.
My suspicions that we're not going to see any fixes for this for months however seem to be confirming themselves. Apparently the bug hunters don't understand PI enough so that the nice instructions and pictures that I posted on my blog and sent to them were insufficient to setup a test to try to reproduce the bug... Face-palm!!!!
Meanwhile the wholesale destruction of PI materials by CCP continues unabated.
|
| |
|