Loren Johnson
- lorenmj
- lorenmj
- America/Chicago
- Yes
- No
MC-17630
For clarity, let me rewrite this:
Now:
This bug appears when a large number of zombies cannot successfully calculate a safe path to their target. Severe lag in either single or multiplayer results as well as time moving jumping forward/backward. The only work-around is to kill the zombies, create a path for them, or switch to peaceful mode. Avoiding the night by sleeping may also work. This is especially noticeable around villages due to the number of zombie that appear and their intended targets (villagers) not being reachable.
Was:
While protecting a village (building walls and putting torches up) a zombie siege event occurs, very often, causing extreme lag.
Could be conflicting code between zombie generation and villager reaction exacerbating each other. The more villagers, the more zombies, the more lag.
----------- From
MC-18865----------------------------------------------------------------I am not sure if all these factors are actually relevant, but here we go:
/gamerule mobGriefing false
Hard difficulty
Villagers inside a house
Zombies trying to break in (difficulty is set to hard, so they would try to break the door, but /gamerule doesn't let them)These things MIGHT be related to the issue. I get really bad delay on picking items, killing mobs and even brewing.
NOTE: this is on SINGLE PLAYER, clearly an internal server problem.
NOTE2: when I kill all the zombies, the delay is gone. There is probably something wrong with the new zombie system as well.
Note that I did upgrade to 13w24a. During testing, just before I did the crash report, a small Zombie Siege was occuring with only a few zombies. When I originally reported this issue there tended to be 10-20 zombies on Easy, so that seems to have changed with the new update. That said, the lag is still occuring even with only a few zombies.
I will set the level to normal and try again on the next night and submit a new crash report to see if anything changes.
On normal level now. I counted at least 12 zombies involved in the siege. Only way to get anything done is kill the zombie or wait excruciatingly long for the sun to rise.
Here's the report:
On normal level now. I counted at least 12 zombies involved in the siege. Only way to get anything done is kill the zombie or wait excruciatingly long for the sun to rise.
Here's the report:
Block destruction is where it is most noticeable. Also if you watch the moon, it moves, then glitches back a few seconds, then moves forward, then glitches back a few seconds. It is unplayable. You pretty much have to SLOWLY kill the zombies or just AFK until the sun comes up.
I get that message on servers where the time on the server is off. A quick ntpdate sync and it goes away, or configure NTP. Not sure what you'd do on a windows server since I don't use it, but probably similar.
The dev version is a server, with 1 client: YOU! So probably the time on your workstation is off. Use a network time server and you never have to set it again.
You are right. I noticed I was getting these too last night during a lag/zombie siege event.
I haven't tested the village again in a while, and I still don't get the block lag, but the timing message is now coming up more than normal. I can always feel when it is happening, then I look over to the console and see that it has happened.
Perhaps this is more related to a lighting recalculation or gravity? You might try the dump and submit it here when this happening.
For me it seems related to the new zombie code (other zombies are attracted when one is attacked and villager reaction to a siege). I didn't have an incredible number of zombies, but everything else happens: time goes off, the moon snaps back. One time a villager just spun in a circle at incredible speed. I placed a block near him and he stopped and ran inside. Seems definitely related to the new zombie code, but probably more in relation to villages and block placement during a siege.
I'm at Cisco Live this week. Not sure if I will get time to test this, but I will try ASAP with the pre-release and see if any other bug fixes fixed this. I saw nothing in the release notes that would have affected this bug, but you never know how a bug fix might cascade and 'fix' some other issue. More to follow when I can get a moment to check the right conditions. I have a map with 4 villages somewhat near each other so I can test this phenomenon fairly easily.
more to follow..
agreed. I just confirmed in my save. The lag is crazy. time can't keep up. the moon snaps back. zombies multiply, villagers go crazy. Still there in 1.6 prerelease.
I notice someone already updated the 1.6.1. sorry, been at cisco live 2013 in orlando this week - at the same location as Minecon will be this year ironically.
In my testing zombie hordes occasionally got as large as 10 or more and was well past the midnight timeframe.
Although the lag would occur even at times that only a few zombies (<5) showed up.
Not sure how the peaceful mode plays into this since no mobs were generated when I play on peaceful. The bug appears when zombies are generating and attacking villagers, so those conditions must be met. No lag occurs when I am on peaceful in the middle of village at any time during the night. I tried twice, two nights with two different villages on peaceful. No lag was experienced.
I can't vote since I reported it, so please all commenters/watchers, vote!
As a work around you can bring it up in peaceful, kill the zombies, then restart in normal.
? maybe??
My favorite server qt.qubetubers.com just upgraded to 1.6 (waiting for mobcatcher to be released). Finally they upgraded. I warned the mods about the zombie issue. A day later they disabled zombies altogether until the issue is fixed. I encouraged them to all vote for this bug.
Briefly my snapshots upgraded to what it called 1.6.4. I was excited to check for the existence of the bug and it appeared.
Later, my snapshot downloader reverted it to 13w38a. This also was affected by the bug. (code upload/naming snafu?)
Either way both versions were affected in my testing. If someone could confirm that would be great. I will leave the affected versions with the 1.6.4 and 13w38a notation.
13w38c showing some promise, however, I don't believe the actual bug is fixed. I got a few ticks of the moon snapping back and also the "system can't keep up message."
Can anyone confirm?
Pic of my setup.
I'm using an Intel Mac Book Air with:
1.8 GHz Intel Core i7
4GB 1333 MHz DDR3
In 1.6.2:
I put a villager in glass blocks (unreachable) and put about 30 zombies in a tank near it. They would try to get to him, but fail. It is only a small area, so no lag. As soon as I allow the zombies out of their tank and (Seemingly) another path out in the world, the lag hits.
I did exactly the same in 13w38c: result no log noticeable.
I am trying to being a 1.7.2 server online, having hardware issues. I am hoping to test 1.7.2 on a server this week if I can get past the issues. I'll post results when I can.
Finally able to setup a server, found a village. While the overall lag is improved during zombie/village interacting events I still get heavy lag. It isn't bad enough to take down the server and I haven't confirmed with a number of people logged in. I was the only one on and did experience heavy lag. The server is running 1.7.2 with no mods or plugins.
At night the CPU hangs around 18-25% usage spikes, but during a minecraft day, after the zombies have been killed it pegs at 16-19%. This indicates some optimizations, but the original bug has yet to be corrected.
Dell SC 1425: 4gb ram, Quad Intel(R) Xeon(TM) CPU 3.00GHz processors.
I have tested the issue. With a fast CPU it is not as prominent, but there is strong evidence to suggest it has been addressed.
See the video for my testing.
YouTubeVideo
I could not duplicate the exact time-lag effects even on older versions. I've upgraded computers and a whole lot has happened since then. I know it was most prominent on servers.
For a completely accurate test I need to duplicate the problem on the previous version, but as you say, that has been difficult to do.
I've been searching for the exact test scenario but haven't come up with it.
Ah! I created the test incorrectly.
Here is likely a more accurate test:
old video
I am out of time for today. I will test soon.
I still cannot duplicate the issue in the older versions. Was the bug in the launcher?
I find that strange. Not that I'm sad about the bug being fixed.
Confirmed on 1.12.1
Same on RHEL 7.2.
It is unwise to run any service as administrator. Not a solution to your problem but you might try running instances as normal users who will store logs in different locations. A thought anyway.