BigCahunaBurger

I've got a budget to buy a new server and/or upgrade my old ones for my farm. Any recommendations for hardware? I was looking at the Dell PowerEdge R720 which is around $700 for 200GB of ram and 40MB total of cache (not sure if it's L3 or higher) - xeon e5-2670. I'm not sure though if it's a bit overpriced, and if the CPU bottleneck is going to make the ram mostly useless. I can already get 250GB of ram on my proliant dl360 g7 for around $300 if I'm so inclined, but the CPU bottleneck seems to be the biggest problem and I'm not sure 40mb of cache is gonna be enough What server would you recommend or build out for up to $2,000 for stealth mode farming?

Hey, I was just thinking of picking one of these up. Actually I was considering the R720. What are your thoughts on the Dell PowerEdge R720? About $700 for 200gb ram and 40mb cache total (not sure what layer) <s>Honestly Kramnik I used to daily drive a proliant DL360 G7 (now I made it headless to reduce fanspin) and running 10 in stealth mode which seems to use about half the capacity of the xeon x5650 (I don't like running all the cores because it's too loud) only draws about 35-50 watts (out of around 500 that I think the server can comfortably pull with a 750 watt limit -- IIRC generally you don't wanna test the psu limit unless you fancy a blown cap or dirty electricity). Back when I used to use it with a GPU it typically only drew 3 watts on idle and about 75 on max load (e.g., compiling a gentoo kernel).</s> EDIT: I'm an idiot I'm glad I posted this cause it made me second guess myself. I was using powertop which was not giving an accurate view of power usage at all. I just finished configuring ilo and I'm using more like 135 watts. So IMO I'm definitely gonna upgrade my server since it's using about as much as OP EDIT: I'm an idiot again because I'm basing the 135 watts off of initial power spike. I'll leave it on a while and see what it averages to Now a proliant isn't exactly the most beast of a server but from what I've seen / read servers usually draw substantially less than gaming PCs Posted a more general thread about hardware if you're interested OP

Thanks Chris, I guess what I'm wondering though is whether you can use the -bot parameter for more than one bot? (So one client for many bots) Like java -jar osbot.jar -bot user1:pass1 user2:pass2 user3:pass3 etc RIP to Zyzz king

Thanks Khaleesi, I appreciate this information. I'll setup my onLoop as you described A few questions 1. Are there any posts I read or info on when to break the loop? I generally understand it should be broken when it has finished its function 2. Would you mind giving an example of what you mean here?

Thanks for your perspective. I guess from my POV I'm trying to make scripts that have very small footprints and easy to handle / easy to debug memory behavior. The goal of my scripts is to be able to run as close to a hundred of them as possible on a given piece of hardware. Of course you can make good scripts that execute a variety of functions and especially I imagine for people who want to level their account or are consuming the scripts as sort of like end users this makes a lot more sense than fragmented tiny scripts that require a manager. Feel free to correct me if I'm wrong about something You're right about case switches from a fundamentals perspective (how the compiler interprets them) but for some reason I just can't get them to behave the way I want in osbot. Explv seems to use case switches with getConfigs which makes a lot of sense but doesn't seem to work for my use case (maybe I'm just not thinking of the algorithms in the right way) I ran into a number of problems with while loops, similar to this and some other threads. I appreciate your distinction between behavior related loops and search related loops, since it seems behavior related loops are the main problem

Making stuff static should mean it won't be garbage collected, which should reduce CPU utilization. I'll compare and contrast. I think pathwalking (not webwalking), preventing unnecessary looping (ensuring the main loop does not re-execute prematurely) combined with static variables / objects should reduce the CPU a lot.

Here's a list of things I've learned to do. I don't fully understand why you should do them, only that life is easier if you do: 1. Don't use any loops (for, while, etc) - all looping should be in the main osbot loop in the main class 2. Use if/then, not case statements 3. Keep all conditions in the main loop 4. If your use-case is RAM and CPU sensitive, use path walk, not webwalk 5. Do not use recursion. Java is not designed for recursion. If you must use recursion clojure should work in theory 6. The beginning of any action should be evaluated and started from the main osbot loop 7. One script for one task, use a manager for multiple tasks

OK, I think I figured out one of the biggest problems here for overuse of the CPU. I was using recursive logic. I'm honestly more of a LISP than a Java guy. I didn't realize that recursion is totally discouraged in Java. I can use it in Clojure and possibly export the clojure jar to osbot, but that is unnecessary work at this level

actually to be honest - I think RAM is fine. I probably want to prioritize high ram and low GC. My cpu is probably going to be the bottleneck, I can get up to like 250gb of ram but I'll be bottlenecking with like 24mb of l3 cache

For farming https://osbot.org/forum/topic/127346-osbot-memory-usage/ Alek talked about max heap size If I want to limit the heap size, are there any recommended levels? Alternatively, it's pretty easy to upgrade my ram to like 250GB but I'm stuck at 24mb of cache regardless of cpu upgrades. Is there a way to increase my heap size and reduce my garbage collection? I'm planning on removing webwalking from my scripts entirely and replacing it with path. Webwalking seems to use too much ram for a serious farm Is there a way to run multiple bots in one cli? That way I don't need multiple separate instances of osbot open (I'm assuming this will use more ram) I realize I tried to do way too much with one script and will now segment it into multiple small scripts that are as tiny as possible, with only one if/then case handling logic in the main loop and only a few if/then conditions per script. The complex script executes fine but at scale seems to be unnecessarily large and likely is hogging more cpu and ram than need be. The logic for handling progressions or changes in the script seems better suited by just exiting the script at the point it should change and starting a new script I am already using the nointerfaces and low cpu cli commands What are some recommended refactoring tips for scripts to minimize ram and cpu usage? Any way I can check to see how big the heap should be with my script? I think I'm thinking of using a memory debugger I read looping more slowly can reduce usage also? How to reduce the loop to once per minute?

Just wanted to note that for me, getting the script to wait for the first VM to close requires using the debug parameter

Ohh okay. Yeah let me see if that works. It should...

I tried with -bot -script twoScript:"username;pass" -script oneScript:"username;pass" as well as "-bot -script oneScript:"username;pass" twoScript:"username;pass" but no luck. I may be formatting it incorrectly. Perhaps you can share your command line format Kramnik, for the record I'm exiting the script on onExit with System.exit(), which is probably not correct.

Nice, so just use the -script flag two times? Or two sets of parameters for one -script flag?

Is there a way to launch a new script when an old one terminates? I'm thinking on the onexit() function. But, any way to do this with the osbot Cli is fine too. I've checked the API but can't find it

LOL, my bad.

I'm guessing a lot of bot detection has to do with the bytecode that is being ran. I am curious if clojure bytecode would run undetected, and if bytecode obfusction performed on public script jars would change ban detection rate. But that's just a theory a GAME theory

This is odd, I've only done f2p so far so I have no experience doing this. What is the abuse score for your IPs? (you could check here, among other places: https://www.abuseipdb.com/) The common theme seems to be your accounts. Maybe the account creation process you are using is getting flagged? If you would be able to share your process without leaking any sensitive info (either here or on a PM), it might help other people if more sensitive flagging procedures are being implemented

Welcome

Video is gone