scoobymad555
Comments
bug STILL present in 46a update - single player on server, stood stationary next to a waterfall also with a lava fall nearby, cpu load at 40%.
come on guys, this is getting stupid now.
To reiterate, this bug was first identified in version 42a and is still present through versions to current one 45a. It's also demonstrated on multiple server o/s platforms with multiple java configurations. Modifying launch parameters makes little to no difference and neither does varying between launching from cli or via a management application such as mcma or mineos.
I appreciate that people are busy and have plenty of things to work on but can we please have an update or at least an acknowledgement that this bug is being looked or has been flagged for further attention? At this point, it literally is a server killer if player base exceeds much more than 10-15 users and as such renders further development of planned projects or developments on the platform essentially pointless!
Brief update - added a few more systems on the world I'm building at the moment and saw my cpu load creeping up again. Having run around under my spawn area and filled in around a dozen large (drop and route) water and lava falls I've clawed about 10% back though. This is probably going un-noticed in single player and lightly loaded servers but any server with 10+ players running around different regions is basically facing a game killer.
Agreed with Ian - lava that's already routed itself i.e. a drop from high level that's already at the floor so, it's displaying a solid block just with animation. From my testing it's also a problem with water on my systems but, it's not always consistent - sometimes it's lava that causes the cpu load and other times it's the water that causes the problem. The only absolute I do have is that when in the overworld, should I load an area with cave systems that have both lava and water I see the cpu load fly up. Removing the source blocks of lava and water in the area see's the cpu loads drop back down to acceptable levels.
Confirmed on my private server. Basic dual core with 8gb ram and sata drives running ubuntu, jar launched with OpenJdk.
Started looking for the cause in snapshot 42a - appears to still be present in current snapshot. Lava seems to have a worse effect than water 'generally' but both cause significant cpu load increases. Currently developing a new world/map which typically sits at a stable 15% load with a single player and in-game clock driven systems running. Ram use is typically around 3.5Gb with disk i/o being negligible barring the odd data burst. As soon as chunks with FLOWING lava or water are within player range cpu loads start to fluctuate and moving closer to them results in higher and higher loads. Have managed to flatline cpu load at 200% repeatedly on demand.
Specifically with lava, it's flow 'range' also appears to be unpredictable - unable to replicate on demand but on two different 'live' server worlds I've found what appear to be naturally generated lava source blocks flowing down 'steps' with lengths over 50 blocks.
Short drops of either i.e. 2 or 3 block drops with liquid spread onto flat surface does not appear to cause significant impact although inconsistently does see marginal cpu load creep of a few percent. Primary issue appears to be large drops and it's subsequent 'pathing' / 'route finding' on it's way down.
Sorry for delay - Doesn't seem to be consistent in character. Hasn't been doing it in the new mac client. Still altering characters in the new client but different ones (speech marks) and only when copied & pasted so different report required really. Mark as closed.
As Ian above,
Windows & Mac clients, CentOs Server on 41b.
Both End and Nether Portals do not place correctly.
All game modes affected.
M/C ver : 1.9 full release, currently allocated 20gb ram (increase from 8gb seems to have helped)
O/S : tried on CentOs 7 and Ubuntu 14.04 LTS
Java : tried with openJDK 7 & 8 and oracle 7, 8 & 9(dev)
Box Spec :
xeon 1225 v3 QC @ 3.6 Gig
256 Gig Ssd raid
Gig Network
Managed by : tried Mcma, MineOs & cli launch
Same errors as listed above - essentially anything that's able to act as a 'vehicle' for the player capable of moving fast : elytra, boats, horses (including skeleton) and donkeys - minecarts don't appear to cause an issue. In some cases our logs are showing horses with no passenger also causing the error.
Errors can be generated at any time - no obvious other activity / trigger event happening at the same time.
Typically the errors become significantly worse and more frequent as number of users increases. Have determined currently that maximum users seems to be around 30 before causing the server instance to crash/terminate.