PDA

View Full Version : Server rollback?



BuddyLee.
11-02-15, 05:51 AM
Im 110% sure the server has been rolled back somehow. Im missing a whole bunch of dinos and equipment.

BL

MeenaSan
11-02-15, 10:46 AM
Hey Buddy,

Unfortunately last night while myself, Vis, Bandit, as well as 3 other brand new players were on the server it was updated. It was two patches behind. When we all logged back in, the server had gone back exactly 2 days (Saturday and Sunday were lost). It was not determined as to if this was due to an error by the admin or a function of the update. As you know, occasionally when updates are implemented you are rolled back sometimes, usually only an hour or so. We were all in shock when we logged back in.

Vis, Geek, and I specifically had spent probably ALL day Sunday converting one of our building to metal and had tamed 2 dinos. All which were lost. The 3 new players kept their levels at least but logged in with nothing.

I don't even know what to say really, to say we were bummed is an understatement...the wind is completely gone from our sails.

Warprosper
11-02-15, 11:03 AM
I'm not sure what happened as I'm not the one who updated it, but what I can tell you is this. We all update the server the same way. We stop the process, run the updater, and then start the process. We don't replace db files or roll back db files. What happened here was most likely the fault of the patch and not the admins updating the server.

I hate that this happened, but with anything early access, we run the risk of something like this. We could attempt a roll back, but from my experience, this has caused more issues than it solved.

Kanati
11-02-15, 02:02 PM
I believe I know what happened and it was likely admin error. Though not my own.

LAM has a known bug. That is the software that I wrote to monitor our servers and keep them running/restart them if needed. That known bug is that if LAM is closed on the server (which it never should be) or if a new process is added to it, it will for some processes, start a second instance of that process. At some point on friday someone closed LAM and I believe I even remember who and when and we decided to leave LAM off at the time so we wouldn't run into this problem.

When that happens, EVERYTHING LAM manages needs to be shut down. Then you start LAM up and it starts all services as normal. Unfortunately that decision wasn't relayed to everyone so someone got on the server and started LAM up thinking that was the thing to do. When that happened, two instances of Procon were started... two instances of Rust were started... Two instances of ARK were started and about 20 instances of Teamspeak were started. Except for using resources, no other game has a problem with that. Most will just go "I can't bind to that port" and be done with it. ARK, though appears to be somewhat of a rogue in that regard. Or so my theory goes.

Here's what I *think* likely happened: Once the second instance started it probably sat there in the same state the other one was in. A duplicate of the first instance. The other instance kept on trucking along and players played their game like normal. Then the request for an update came. When I got on, I noticed that the dupes were running and decided to fix everything at once. Again, this is not a problem with anything BUT ARK it appears. I closed everything down. But here's where the chaos comes in. I closed down the instance of ARK that people had been playing on. It wrote out it's game state. I then closed down the duplicate (you can't tell which is which really without studying the logs, but since it wasn't known at the time to be problematic I just shut them down in the order I saw tham). IT then wrote out ITS game state, which was basically an image of friday night or saturday morning, overwriting what the other instance had already written out. Effectively rolling shit back 2 days.

Now that it's KNOWN, we can probably avoid it in the future. As for what has been done... there's not much that CAN be done. It is what it is unfortunately. Educating the other High's on this issue is obviously needed (again in some cases). And perhaps and "YOU ARE ABOUT TO CLOSE LAM! ARE YOU GODDAMN SURE YOU WANT TO DO THAT!?" prompting needs to be added to the program to prevent accidental closure of the tool.

CleverGeek
11-02-15, 03:37 PM
Thank you for looking into that!

It's hard not to repeat issues when you don't know the root of them. We thought it might have something to do with multiple updates being required, either way though we will keep on top of requesting updates in a timely manner.

I am glad to hear that there was something causing excessive resource usage as well, because within the time period that was rolled back the lag and play-ability was horrible. Now that you have corrected the issue it should help out the performance issues as well.

Again, thanks for the help!

visuvius
11-02-15, 03:40 PM
Yeah as soon as he mentioned "two instances of ARK" I assumed that is what was causing the worse than usual lag on Sunday.

Kanati
11-02-15, 03:55 PM
Because some people's noses got bent a bit out of shape from my explanation... Let me clarify...

MY software has a bug. MY software is able to be closed inadvertantly. MY sofware can be re-opened and cause Very Bad Things (tm) to happen if things are not started up in a particular way.

So blame me. Only me. All of me. In every single way possible for the problem ARK experienced with the rollback. If the bug wasn't present, it wouldn't have happened. I will be looking into and fixing that bug in the near future.

Option hunter
11-02-15, 03:57 PM
Glad to know we can avoid this problem in the future. The three new people were me and my friends, and we're still going to play here, so there's no harm done there. :)

BuddyLee.
11-02-15, 07:07 PM
Im butt hurt. I spent alot of time playing this weekend.

Butty Lee