ShardManager with 12000 ms, Any ideas to solve this ?

[megaman963]

let's put a thesis, if SR_ShardManger has job on:
1. SQL DB
2. Scheduling : Events, Spwans, Buffs .. etc
3. FTP for crest
4. Establish Billing server
5. Connecting to SR_GameServer
6. Connecting to AgentServer

It will do:
1. Inserting-Deleteing-Updating rows, Execute Stored Procuders, Reading Tables
2. Sending Packets to SR_GameServer for schedules
3. Uploading encrypted bmp files into FTP
4. Checking with billing server
5. Send packets to SR_GameServer and AgentServer
6. Creating Dumps for error handling or anything

Check out which of these has relation with your high ms ...

maybe with billing server ??

[Tazdingo]

let's put a thesis, if SR_ShardManger has job on:
1. SQL DB
2. Scheduling : Events, Spwans, Buffs .. etc
3. FTP for crest
4. Establish Billing server
5. Connecting to SR_GameServer
6. Connecting to AgentServer

It will do:
1. Inserting-Deleteing-Updating rows, Execute Stored Procuders, Reading Tables
2. Sending Packets to SR_GameServer for schedules
3. Uploading encrypted bmp files into FTP
4. Checking with billing server
5. Send packets to SR_GameServer and AgentServer
6. Creating Dumps for error handling or anything

Check out which of these has relation with your high ms ...

maybe with billing server ??

1- SQL DB Maybe
2- Scheduling i think this is the problem. But what can i do ? I can delete all the data
3- FTP crest (Not i already test it)
4- Billing server (i will test that)
5-6- The relation with agent and gameserver is OK

Thanks for answer (Y)

[Extornia]

So 2 quad processors with each 8 HT totalling to 16 is not enough?

Are you fucking retarded?

[theross]

*cough* suggestion - sql 2000 *cough*
had the problems on a customers server as well.
if you have any "disk activity tool" you'll see that at the high rate the hdd is writing some stuff into the shard db.
the shard manager collects some changes, and at a certain time transfers the whole bunch to the sql server, which causes quite some disk activity. the bigger the db is, the more time it takes.

there weren't any big things we could do.
the client installed a ssd as database drive, and it helped a tiny little bit.
we figured out that there was a consistency problem in the database causing the huge disk activity.
After fixing those consistency errors it went pretty well.
ps: it was only an i7 cpu with 16gb of ram holding 700+ people as well ( 2 machines). we shall not exaggerate with the system requirements of the sro server files.

personal opinion - it has nothing to do with ram or cpu (wether i studied engineering, or i have several years of experience in administrating server systems). people who claim that dual or quad-xeons are better, are just..... different in my point of view. those kind of people also would buy an old vw polo and "tune" it instead of buying a reliable car :)

Anyway, maybe these infos will get you further with your problem.
Greetings,
SilkBotter

[Blackdiana]

i think you not use 2 server.. you have 1 server, and second machine runing in hyper-v or virtual machine... This is bad thing.. if want better and lag free server, you need 2 dedicated server in same host...

[Tazdingo]

i think you not use 2 server.. you have 1 server, and second machine runing in hyper-v or virtual machine... This is bad thing.. if want better and lag free server, you need 2 dedicated server in same host...

Sorry but you 're wrong.
But i apreciate all the answers, thanks.

"theross" knows about the matter is the only one who answer something usefull.

[xxNukertube]

Processor

[Tazdingo]

Processor

Hard disk

[TryToMakeServer]

You have custom queries, which overloads the shardmanager when excuting a original procedure.
I had the same problem with 26k ms.
I had added some extra queries in _ADDNEWCHAR query which overloaded the shardmanager when anyone created a new char.

Just try to find another procedure which wont get excuted everytime.