SQL Errors

[desrat 4]

Ever since I set up EU16 I've been getting errors like the ones you see at the bottom of this post from hive.exe and similar from hiveAuth.exe, now it seems that when these errors start occurring the app never recovers (every subsequent entry is an error).

My question is Can you detect when these errors start occurring somehow? I run my own restart tool, and restart hiveAuth.exe every 15 minutes as it seems to be the most susceptible to these errors, now can I do that with hive.exe too? I was given the impression that if hive.exe needs to be restarted then so does the server (arma2oaserver.exe).

# ERR: SQLException in .\hive.cpp# LINE359
# ERR: MySQL server has gone away (MySQL error code: 2006, SQLState: HY000 )
Sent...201

[Mojo (DayZ) 66]

Oh, we all get those. 2006 is a "MySQL server has gone away" error. While the hives can be quite verbose, I have no idea if there's a debug or log flag to actually save that verbosity to an output.

If there's not, I'm throwing my voice behind hive logging. (And if there is, please, God, tell me what the hell the parameter is).

[desrat 4]

its annoying sometimes it can go the full 6hrs between restarts and others it can be minutes - definitely needs some way to monitor - or better yet a fallover procedure in hive itself to shutdown (so it can be detected easier)

[Mojo (DayZ) 66]

its annoying sometimes it can go the full 6hrs between restarts and others it can be minutes - definitely needs some way to monitor - or better yet a fallover procedure in hive itself to shutdown (so it can be detected easier)

Indeed' date=' you hit upon the other thing I wanted to see. A way to force fail state for automatic recovery in the parameters. If one isn't well-versed in computer administration (be it server or PC or what have you), forced fail states can often be catastrophic and far more annoying than error generation. On the other hand, if one [i']is sufficiently experienced with what they're doing, forcing a program to fail, especially when entering an otherwise benign error-reporting state, is sometimes ideal (and far more graceful).

For example, I wouldn't want hiveauth to force fail if it got a single Error 2006. There could be many reasons for that, some of those reasons causing hiveauth to immediately refresh its connection could compound. On the other hand, terminate application after X minutes/increments of errors/retries would be a very nice option.

[Vipeax 318]

I have this odd feeling that this server is not managed by the team/self-managed as I fixed that a while ago already....

[desrat 4]

Wanna shed some light on the subject then instead of holding on to said information? Rocket has already said he prefers hosts to manage their own servers so if there is a fix for this it needs to be made public knowledge...

[Vipeax 318]

Not a big fan of putting these files in the open. While it is indeed preferred to have hosts, mainly the ones that consider themselves knowledgeable, manage stuff on their own, they are also the ones harder to reach for updates, and all be it hard to imagine, I have more to do in the hospital than try to hunt down 300 server hosts ;). I'm quite easy to find on (the official) IRC most of the time, thus generally speaking it's handy to come by and ask about it once in a while.

[Zarlak 339]

Oh' date=' we all get those. 2006 is a "MySQL server has gone away" error. While the hives can be quite verbose, I have no idea if there's a debug or log flag to actually save that verbosity to an output.

If there's not, I'm throwing my voice behind hive logging. (And if there is, please, God, tell me what the hell the parameter is).

[/quote']

When you say Hive Logging, do you mean being able to read .txt file with all of the Hive entries stored in it? Or something along those lines? If so this should be a huge no. No matter how helpful it would be.

[Mojo (DayZ) 66]

Oh' date=' we all get those. 2006 is a "MySQL server has gone away" error. While the hives can be quite verbose, I have no idea if there's a debug or log flag to actually save that verbosity to an output.

If there's not, I'm throwing my voice behind hive logging. (And if there is, please, God, tell me what the hell the parameter is).

[/quote']

When you say Hive Logging, do you mean being able to read .txt file with all of the Hive entries stored in it? Or something along those lines? If so this should be a huge no. No matter how helpful it would be.

Yep, that's basically exactly what I mean. Every chatty little statement hive or hiveauth generates is dumped to a text file. If it's a concern of tracking players, I can already watch things like the class names of their inventory show up in hive. Trust me, if I wanted to monitor particular players, I would first need to divine who they are, and even then it's much more useful (cheating-wise) to do it in real-time than through a log from several hours ago. If it's a space concern, if you're the kind of host that likes to obsessively log things, you're the kind of host that can ren incremental log files and del the oldest ones through batch scripting.

I like to know what goes into and out of the expensive stuff I'm legally and financially responsible for, and maintain it at its highest possible operating capacity, either through active intervention or regular feedback to the application developers. That's just ol' control-freak me.

[Vipeax 318]

What is it that you want to log in the first place? I guess I can imagine it being useful to log the time and kind of MySQL error, but why would you want to log every kind of query being run? Is it that interesting to know an Object-Stream occurred, which is as it's supposed to be?

I've been working on fixing the memory leaks, re-establishing connections upon loss etc on newer versions of Hive, adding a full time logging feature is something that I consider only a waste of memory, IO and even time as time is already a bitch as it is.

[Mojo (DayZ) 66]

What is it that you want to log in the first place? I guess I can imagine it being useful to log the time and kind of MySQL error' date=' but why would you want to log every kind of query being run? Is it that interesting to know an Object-Stream occurred, which is as it's supposed to be?

I've been working on fixing the memory leaks, re-establishing connections upon loss etc on newer versions of Hive, adding a full time logging feature is something that I consider only a waste of memory, IO and even time as time is already a bitch as it is.

[/quote']

No, it's not interesting to know when an object-stream occurred. It's not a matter of interest, it's a matter of raw data. I despise when event loggers decide for me what events are important or not. If I want to filter the raw data, I'll decide how to parse/plot the information, not the program. At its most tedious this means isolating several samples of unwanted events and filtering out all instances of like entries with regex in notepad++

[Zarlak 339]

Oh' date=' we all get those. 2006 is a "MySQL server has gone away" error. While the hives can be quite verbose, I have no idea if there's a debug or log flag to actually save that verbosity to an output.

If there's not, I'm throwing my voice behind hive logging. (And if there is, please, God, tell me what the hell the parameter is).

[/quote']

When you say Hive Logging, do you mean being able to read .txt file with all of the Hive entries stored in it? Or something along those lines? If so this should be a huge no. No matter how helpful it would be.

Yep, that's basically exactly what I mean. Every chatty little statement hive or hiveauth generates is dumped to a text file. If it's a concern of tracking players, I can already watch things like the class names of their inventory show up in hive. Trust me, if I wanted to monitor particular players, I would first need to divine who they are, and even then it's much more useful (cheating-wise) to do it in real-time than through a log from several hours ago. If it's a space concern, if you're the kind of host that likes to obsessively log things, you're the kind of host that can ren incremental log files and del the oldest ones through batch scripting.

I like to know what goes into and out of the expensive stuff I'm legally and financially responsible for, and maintain it at its highest possible operating capacity, either through active intervention or regular feedback to the application developers. That's just ol' control-freak me.

Yeah, I get what you mean. It just would be a nono for me. Hive logs everything but it scrolls quite fast so you can miss a lot of stuff. When someone interacts with a vehicle for example and saves it it gives the map co-ordinates. Now just imagine abusing admins with the log file seeing where the jeeps and hueys are and heading over to steal them. Or noting there was a silenced weapon put into a tent at X Y and heads over to find a base camp with 5 tents full of stuff.

I know it happens in Hive already, but at least you can miss it if you're not looking at Hive constantly. Giving someone the whole log with everything in it would spoil some servers no doubt.

[Vipeax 318]

Give me a poke on IRC and explain what you'd like to see/added as a feature. Can at least consider it :).

[Mojo (DayZ) 66]

Yeah' date=' I get what you mean. It just would be a nono for me. Hive logs everything but it scrolls quite fast so you can miss a lot of stuff. When someone interacts with a vehicle for example and saves it it gives the map co-ordinates. Now just imagine abusing admins with the log file seeing where the jeeps and hueys are and heading over to steal them. Or noting there was a silenced weapon put into a tent at X Y and heads over to find a base camp with 5 tents full of stuff.

I know it happens in Hive already, but at least you can miss it if you're not looking at Hive constantly. Giving someone the whole log with everything in it would spoil some servers no doubt.

[/quote']

And trust me, I understand your point as well. On that point it might be better to flag administrator characters and run an algorithm to detect if they are in the vicinity of such an object unflagged as owned or previously operated by them to a significant degree more often than the average player base and begin more active monitoring procedures if they raise red flags, instituting bans and blacklists if such a case occurs.

As it stands, given that the hives are already subject to that to a degree, and that with the mission files available to the hosts as they are, the possibility for abuses equal to or exceeding that already exist and that the focus should be on mitigating the greatest damage.

With log files, consider that an administrator could filter for eg: object stream events which occur too frequently for such objects in too small a radius in too short a time. For example, within POS X,Y + DIST U,V where POS is a fixed location and DIST is a place relative to that, three or four calls on class NVGoggles or M136 within only a few minutes either means a North-West Airfield miracle or endemic player item duping.

At this juncture, it being an alpha and that being the stage at which you want to weed out as much of this behaviour as possible, I'd argue the merits of this kind of exploit/bug/glitch/abuse detection, especially when suspect groups are identified as being on the server, would solicit far more active abuse quashing than it would active abuse causing.

If and when the cheating has been significantly mitigated to an acceptable minimum, then the logging can be removed if desired.