Jump to content

Ixel

Pathfinder
  • Content Count

    41
  • Joined

  • Last visited

Community Reputation

3 Neutral

About Ixel

  • Rank
    Landlubber

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. @Jatheish I'm still waiting to know whether unofficial servers will get an updated grid editor before the expected release date in regards to potential new content coming in the major patch next month? I've asked this in two different posts but I didn't see a reply from either you or @Dollie.
  2. @Dollie Will unofficial servers have the option of participating as a PTR server, or will this not be possible due to the game client also needing to be updated or for another reason?
  3. Regarding Redis, I'd recommend not running it on Windows Server if you end up having a busy grid. Redis runs much better on Linux and doesn't end up gobbling huge amounts of memory per save it does (if on a busy grid) unlike on Windows Server where it seems to do so. I'm running mine on a virtual machine in Debian, obviously disabled things like transparent huge pages.
  4. I redid my grid recently to an 8x8, after the recent patch. I now have a few empty zones ready for potential new content needing to be manually added. I hope they update the servergrideditor a day or so before the actual patch goes live so as to give time for unofficial server owners to place the new content ready for the live patch. It astounds me that I managed to get it to fit in 128GB RAM but I'm using both memory page combining/KSM and a little memory compression. No noticeable performance impact for players.
  5. A way to override alpha spawns would be nice, unless I've overlooked something.
  6. The Day0 thing is quite puzzling, it reduced my CPU usage on the 5x5 PvP grid from roughly 50% to now roughly 15%. No side effects noticeable. If it wasn't for RAM usage I could host both the 7x7 and 5x5 grids on my 2990WX CPU with the kind of CPU usage I'm now seeing since changing Day0 to a date in 2018. Maybe I can partially use a pagefile/swap, as I think some of the memory used by Atlas is somewhat inactively used (e.g. for storing the island layout or such in memory). I currently use ESXi which has a Windows Server guest virtual machine for the 5x5 grid, ESXi can of course also share memory pages between two different virtual machines if those memory pages appear identical, thus reducing actual memory usage further with minor performance impact (before resorting to a page file). The 7x7, for now, is hosted on a different system but that's not my own hardware so am paying much more than I would if it was on my 2990WX system.
  7. Are you sure it requires a redis wipe? Unless I got lucky I tried it once and just wiped the save file, and it worked. But otherwise yes, it does get stuck without some type of wipe and my suspicion is that it's something to do with the NPCZoneManager (or NPCShipZoneManager I believe) only fetching the setting once, unlike the difficulty settings). The same appears to be happening with whales, I can't change their spawn rate without wiping the save file.
  8. Bump. EDIT: Nevermind, nobody has voted so I guess this forum isn't really for dedicated server owners to report bugs. I'll try reporting it by other means such as Twitter, Github or such.
  9. Assuming anyone official reads this bug report, I setup a test grid since my live one is suffering badly from excessive ship of the damned (which I've now disabled by using a spawn override in the game.ini file, for now). Basically what I believe I've discovered is that the NPCShipNumMult setting works in one direction, upwards. For example, if I set the multiplier to 0.50 in a zone and then decrease it to 0.01, restart and destroywilddinos, they still spawn as if it's a multiplier of 0.50 and not the new lowered rate of 0.01. However, if I do it in the opposite direction and increase it to a number such as 1.00, the change is noticeable. It seems possible to increase it after the first time of running a zone, but decreasing it seems to have no actual change. Changing the difficulty settings however works fine.
  10. Going for a multi socket or a CPU such as the Ryzen 2990WX might work, however if memory bandwidth is insufficient then there will be a bottleneck again.I don't know how much memory bandwidth Atlas tends to consume as I haven't monitored that, I only know how much memory it actually consumes.
  11. In the patch notes I'm wondering why you've restricted this feature to official servers only? I realise that official servers are much busier but still... this could be a useful feature, especially if the duration is customisable, for unofficial servers.
  12. The clock speed isn't the best, but not the worst either. It can probably cope with 25-30 players each zone, if it's a zone with plenty of structures or such, I'd say. I wouldn't suggest going much beyond that per 'populated' zone on that CPU. Like ARK, ATLAS prefers higher clock speed per core as it's more single threaded than multithreaded.
  13. One option to possibly get this kind of concept working on a dedicated server level might be to use 'SetGlobalPause'. While it wouldn't stop RAM consumption, it should still keep the server alive but hopefully in a state where it demands much less from the CPU. You would need something to monitor the number of connected players (once every five seconds maybe?) and to disable or enable 'SetGlobalPause' accordingly via RCON.
  14. I run five zones each on those virtual machines at the moment, however I do get a fair number of players in some zones. I've noticed a little lag at times, but there have been a few other situations that may have caused that such as catfish overspawning in a pond on a specific island (which I worked around by adding a spawn override to the defaultgame.ini). I'm guessing more players in a single zone just needs a higher clock speed per core such as an i7 CPU. Another problem I've been experiencing recently is redis crashing and running out of memory. Might be a Windows specific issue with Redis though, given Redis isn't officially maintained or supported for Windows anymore. I'm setting up a Debian VM for Redis. One other issue I've noticed is that the IP's from netcup seem to take a good while to show up in the server list when searching.
  15. Did anyone using AtlasUpdater manage to get an auto update last night? My three systems with AtlasUpdater didn't seem to acknowledge an update until I restarted AtlasUpdater.
×