Grntrenchman

  • Content Count

    5
  • Joined

  • Last visited

Community Reputation

0 Neutral

About Grntrenchman

  • Rank
    Junior Member
...
  1. I had a script wiping all the workshop-* folders before starting server, as it wouldn't download new versions otherwise. Now only 2 mods (gemcore, insight) download/load. There are no obvious errors. Here are SS of the folders and dedicated_server_mods_setup.lua, and the log is attached. server_log.txt EDIT: So, this works fine in a win10 host instead of xp. also, I caught an error that didn't show up in logs (on xp) in another SS
  2. I was experiencing problems running a dedicated server on CentOS, and have tracked it down to the drive the server is running from having too much free space. I'm not sure where the cutoff will be, but most likely an overflow somewhere in code. Symptoms are databundle reading being skipped, errors about reading free space, lack of saving, and a corrupted world. Here is the thread with more info and troubleshooting. I'm still testing it more in a VM, but have it working otherwise on the same system with the same files using a smaller loop device of 16gb. In a test VM, making the drive very large brings the error about free space, but it still reads datastores. I tried 20tb, but on reflection, my actual server runs on a 36tb drive, so i'll make the VM drive bigger and report back. As a note, the server runs fine in a windows VM accessing the (same) 36tb storage, but reports only 4.2gb of free space (supporting my theory of an overflow, just caught here) EDIT: Tried with a 40tb vhd, still starts after resizing filesystem... not sure about this one. Maybe something specific about the file system?
  3. I saw those also, and they lead me to think we do have a library or code problem. I've clobbered the whole server dir repeatedly with 777 during troubleshooting, so it's definitely Read/Write worthy. Here is a pastebin of ls -laR for the save dir... dontstarve user has read/write to everything, in fact has made most of these files/dirs. at worst, OTHER users don't have access. even better that in those same logs, it confirms access just before the first instance of that error... there's 11.5TB free on that volume, so unless we have a counting overflow problem... shouldn't be an issue. I'm down into the strange stuff, temp disabling selinux etc. EDIT: It is probably a code or library issue. I noticed the logic for detecting space isn't working great in windows either. It shows 4.2GB free for the same volume, with 11.5TB free (capture attached). The working centos server is in a VM with a 50gb drive. I'm going to try a smaller partition for this server on the non-working centos install. EDIT2: That was it. Something is wrong with linux port, can't handle free space, probably an overflow. off to bug reports! I took my linux copy of server, stuck in a 16gb loop img file, mounted it in the same place the dir was for linux, even got my newer save and mods from windows copy, boots fine, loads fine, no more mounting databundles skipped, no more free space errors, it loaded the map instead of generating a new one... good to go. windows is probably using different libs/method for this, and fails.
  4. I've installed dedi through steamcmd in a win10 vm, copied the config/save from the linux copy, tweaked ip settings and scripts for batch instead of cmd, and it starts and runs fine. No new character or anything, loads right up. I've gone back to the linux version, and it made a new save on start, which is corrupted. I caught video of it for you. Again, invisible wall, which is obviously a different biome than shown (water?). I'll blow away the server and start again with steamcmd. I did use that to install it originally. I can start listing the various ways i'm calling things to help. (attach start/update scripts etc) Also, here is the log from that server run. server_log.txt EDIT: I've cleaned up the install and reinstalled the server. it gives an error about the library: then the original symptoms: then it closes. here is the environment and scripts. I called the update.sh, then startdst.sh. I'll look into building proper libcurl 32-bit instead of linking to it. I don't know how to tell if it's the right version, i've had to tweak libs for both ark and factorio at this point. EDIT2: I made libcurl from source with and it compiles no problem. I've made the link to /usr/lib/libcurl-gnutls.so.4 from the one made with ./make install in /usr/local/lib/libcurl.so.4 instead of the one that comes with the system. The symptoms are the same. I'll try unpacking the databundles again. EDIT3: I installed the Dedi on a centos VM I have, and LDD shows wayyy more libs on this system, and it runs properly first try? haven't connected to it yet.. Working install (didn't require unpacking databundles): Broken install:
  5. Running CentOS 8, the server tries to start, but fails. I saw another thread with same issue and no replies... like so: extracting the scripts databundle into data makes it go further but still crash, missing zips from the anim folder. lots of: I extracted the rest of the zips into the data folder, and it starts now. I've just attached a 7z extract to run after steamcmd update to keep the datastores updated. I've made the necessary symlink from libcurl to libcurl-gnutls and that allowed the server to start in the first place. I've got i686 packages of the other requirements installed properly. When I join, it kicks me after a short time (5s to 2m). I'm also encountering lots of pop-back, and maybe even invisible walls. When I first spawned in, it put me in a swamp and I couldn't walk out of it. Second time was on land, and I had a invisible border randomly across some ground. Not sure what to troubleshoot here, so I'm going to attempt to run it through a windows VM (goodbye extra cpu/ram, thanks wandows). I'd still like to help figure out what's up with the linux version if I can, for you guys, and if it's fixable so I can have those extra resources back.