Just want to suggest that estimates for the duration of server checks be overestimated. For example, if it looks like its going to take 30 minutes, tell us the server check is 2 hours. If a server check ends earlier than anticipated then great, the staff are deemed heroes and no one wastes their time waiting way past the server check estimate. Compared that to what happens now more often then not, which is that server checks go beyond their estimate, staff are deemed incompetent, and people's time are wasted waiting beyond the server check estimate. Please save us alot of time and save yourselves from the hate. If there is a legitimate reason why you guys don't do this that actually outweighs the benefits, please share with us.
I believe they're already over-estimating. But sometimes even an over-estimate may prove to be insufficient.
So I wasn't around last night/today when this server check was done, however, I did read all of the Skype chat that occurred amongst the staff during that time. The update had been ready to go, the server was shut off and the update was pushed, followed by the server quickly coming back up. There was an "oh crap" moment that occurred when someone on staff read the shoutbox and noticed a player stating the AP reset had some adverse effect on their stats. Kevin immediately looked at the code, realized he literally had used "-=" where he should have used "+=" and had the server quickly shut down to not only push that change in operator, but to prevent other players from getting screwed over. This just serves as an example to illustrate that while yes, there are times we should have announced a longer downtime, there are likely equally as many times where our estimates were correct, but an issue is found shortly after bringing the server back up. It is impossible for us to test every scenario on our own but we certainly do try to avoid issues showing up with our players. I hope that gives some clarity to the OP's feedback. If not, I'll try to explain better if you have any questions
On top of that ^ the longer downtime was nothing more than waiting on the backup to import. I believe it was simply mysql taking its sweet time with the query and there was absolutely nothing we could do to speed it up. It's like when you are downloading something and it fluctuates from 5 minutes remaining to 2 hours remaining, then down to 1 minute. If it says 30 minutes remaining, then that's all we can go off of... If it says 30 minutes remaining and it takes 45 then that's nothing we could change. Edit : also some people get just as mad for the server being brought up early as some people do for delays. Lose - lose no matter what you do.
Most of the time we are not far off when we give notices in game about when the server goes down for an update. Today was more of an anomaly like John mentioned since we had to fix something quickly, and re-importing a database backup too longer than usual for whatever reason. If we overestimate - like in the recent patch where it finished 25 minutes early - this still annoyed some players as they expect the server to go online at the time specified and were upset that they missed out on the 25 minutes of gameplay. The last thing we want to do is overestimate and then wait out the full duration of our estimate if everything is ready and the server could be started up, since we would obviously prefer to have as little downtime as possible, and I'm sure most of the players prefer that too. Therefore I feel it is best to try and give accurate estimates whenever we need to give estimates, and then extend these estimates if required.
Thanks for the responses. You guys mentioned that some players are dissatisfied with server checks ending earlier then scheduled. I'm doubting that they are near the majority, and I would assume that even among those people many of them would rather have it end earlier then scheduled if the alternative is instead to end later then scheduled. Personally for me being a fm spot hunter and all, I would find it annoying to have server checks end earlier than usual, however as someone who values his time, I would be far more annoyed if the server check instead wasted a bunch of my time. I just don't think that because a minority of players somehow find it an even greater inconvenience for a server check to end early instead of late, that we should equate ending early to be just as bad as ending late. But maybe I'm wrong here and the people being dissatisfied with server checks ending are early are actually close to majority. Is it actually? or is it just "some"?
Yes overestimating might be insufficient sometimes, but it would reduce the frequency of late server checks by a landslide, which I believe is very beneficial.
Ngl when i saw the 6 hour estimate for the CS update, a part of me believed they actually estimated like 20 minutes but said 6 hours just so people wouldnt be mad when they delayed it
As much as we would all like for things to be ready right now. That just doesn't happen unfortunately. They cannot foresee what problems may happen or other things that may delay the server checks. Good example, last night, the first server check was estimated at 10 minutes. It was that long, but because there was a bug, things got delayed and took longer
I think they are just trying to be as straightforward as possible. We expect it to take 15 mins but... it could take more.
Well the thing is that when they tell us they expect to be done in 15 mins, players will also expect them to be done in 15 mins. If I cant expect what the staff is expecting it means I can't trust the staff. It seems server checks have been delayed very consistently for almost 4 years now, at the very least I know it has been for the 1.5 years that I've been here. I think it only makes sense after that much experience to start considering possible delays before announcing the estimate. Think about this way. Let's say I had to meet a friend and my gps tells me it's going to take 1 hour to get there. Now let's say I have a tire that's going flat, low gas, and the gps is leading me through a freeway that has frequent accidents. In this case it would be irresponsible to tell my friend that I expect to be at the location in 1 hour
I would argue that the majority of the server restarts to implement updates since the new source have taken around 15 minutes or less. I don't see any reason to start saying that these updates are going to take say 2 hours when it's likely that they will be done in ~15 minutes as usual.
Doesn't have to be 2 hours that was just an example and I did mention that it was an example. Why not give yourselves an extra 15 mins since there is a good chance you'll need it?
While I agree that overestimating is usually a good practice, it's worth noting that the server went up within 15 minutes after being brought offline. The following delay was, like John previously stated, due to an oversight on our part which only affected a part of the player base. Reimporting the database then took longer than expected even with a slight overestimation so I doubt it would've made any difference in this case.
I find it weird that in a thread where feedback is given to the staff, the only person thanking anyone is the person givig the feedback
I personally never considered simple restarts to unstuck a channel or something to be a server check, but then again I admit I don't know the exact definition of a server check. If as Matt says timely server checks are going to be a "usual" thing from now on then that's great for all of us. However if the trend starts looking like it did in the previous source I would hope that the staff would consider what was discussed here
