Jump to content
  • 0

lag dependent server auto restart


nameless

Suggestion

lag dependent server auto restart

 

server can be programed to check server lag. When heavy lag is detected, it auto save and restarts.

Such as:

if (lag detected (such as, pin> 180 in more than 50% of the user))

    {

    disable initiation of boss arena, missions, ascensions, OSD & element vein defense;

    if (no  boss arena, missions, ascensions, OSD & element vein defense is currently in progress)

        {

        save;

        restart server;

        }

    }

 

additionally, servers can be programed to freeze time for 5 minute upon any restart (as opposed to destroying element vein, OSD, or ending boss arena, missions, and ascensions)

Link to comment
Share on other sites

8 replies to this server topic

Recommended Posts

6 hours ago, YadoKURD said:

First of all that the way of "programming" u typed, its meanless, cause lag happens due its own auto save, the lags that are too much.

I didn't mean the "lag" that happened during saves every 15 minutes. 
 This suggestion meanly prevent server crash before server saves, and whether a server is about to crash is judged by the pin of the users (I believe there is a better way.) I think that will help with, character loss, destruction of element vein, OSD, or abrupt ending of boss arena, missions, and ascensions.

Link to comment
Share on other sites

Just now, Aylana314159 said:

Not sure this would work, since the joining the servers isn't ping limited. 

the point is that once the server detects its in risk of crashing, it prevents initiation of boss arenas and waits for current arenas to end. Once after that the server saves and immediately restarts

in short, the saving time point read upon restart will always be one in which no arena is in progress.

the most common way of losing character is transferring a character: character is downloaded into a server and this server crashes before it can save. Thus character is deleted from cloud, and the save read upon server restart do not include the character.

if there the period of time between any adjacent server save and server restart is always zero, then no transferred item/ character can be lost. (of course, this is what I thought. The real situation depend on the current server protocols)

Link to comment
Share on other sites

20 minutes ago, nameless said:

the point is that once the server detects its in risk of crashing, it prevents initiation of boss arenas and waits for current arenas to end. Once after that the server saves and immediately restarts

in short, the saving time point read upon restart will always be one in which no arena is in progress.

the most common way of losing character is transferring a character: character is downloaded into a server and this server crashes before it can save. Thus character is deleted from cloud, and the save read upon server restart do not include the character.

if there the period of time between any adjacent server save and server restart is always zero, then no transferred item/ character can be lost. (of course, this is what I thought. The real situation depend on the current server protocols)

I get what you're saying, but this would just make the servers constantly restart.

Link to comment
Share on other sites

2 hours ago, Aylana314159 said:

I get what you're saying, but this would just make the servers constantly restart.

Yup, that's why the threshold is important. Ideally, this mechanism should be triggered when and only when the server is unplayable or about to crash.


In the patch notes I saw ark devs have collected server data to "fix stall and crashes". If they did what I thought they did, now they should have an array of data collected from server when it crashes.  (Such like Sever XXX, crash log at sever click XXX, time and date XXXX/XX/XX XX:XX, ………. in the last save player AA is connected with pin BB at location CC, player DD is...….. so and so on, from a lot of servers)

 

Since official servers with the same map are mostly identical to each other software wise, an algorithm can be made to seek correlation between event "sever crash" and certain variables in the crash log. (Just like how an array of personal genomes can be used to correlate event "pathogenic mutation" to certain alleles in genomes, though hopefully their crash logs are not as long as a genome...) 

That being said, if they had good quality crash log, they should be able to generate a pretty accurate prediction of when a sever is crashing. If they don't...… then the pin is the only way I could think of...……..

 

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...