Well /g/?Every IIIT grad can answer this btw.
My watch
>>108826670Have I recently downloaded Cemu or any python libraries with PIP?
>no cron jobsBecause it's a systemd service or something scheduled by the application.
>>108826670Logs. Next
>>108826670I ask Claude because I’m not a ludder
>>108826670i open htop, sort by cpu usage, and stare at it until 3:17am
>>108826780This and also request over time.
>>108826670nothing and decommissioning the machine because I was just told it's not in use
>>108826747Based vibegod
>>108826699kek this
I don't check shit, unused CPU is wasted CPU.
>>108826670I turn off my computer when I go to bed because my best friend deserves to sleep too.
>>108826699>2026>logslmaoooo
>>108826670>Jahir SheikhIm cheking my local ICE department
>>108826670Network connections.
>>108826670my app or the node spikes?
Not my problem, boss
>>108826670flame graph?
>>108826670I reboot the pc
I would first check out the importance of that number 317.
>>108826670>checking logs is a "senior" skill to an Indian
>@claude investigate this issuenext
>>108826670i'm interviewing for backend, not sysadmin. so not my problem.
>>108826670Use elastic observibility or whatever they got, highlight the timeframe and go from there.
>check logs>our retarded logging is too noisy and useless to determine any meaning out of it, despite attempts to have it cleaned up to be useful>nothing is stated broken>close as wontfix
>>108826670Systemd timers
>>108828056Anyone who says logs would end up here. If your app is important enough you know CPU usage jumps at a specific time, you probably have a good APM and opentelemetry anyway.
>ls>lscpu>top>pretends I know what I'm reading>q
>>108827273ThisOP is a poorfag and can't afford a bigger CPU
>>108826670logs/graphs to see if anything else occurs at 3:17am it should easily be visible in whatever monitoring software there is if you use datadog or something there is literally something built in that will check across everything in your entire company for what else has a similar spike at that time whether it's a certain type of request or error or w/e
>>108826670nothing. I'm asleep at that time and I don't care about the company's power bill.
>>108826670Trick question, we use next.js, CPU is always at 100% sir
Install a claw code bitcoin mining agent team running Opus 4.7 to balance out CPU usage. The boss is gonna freak if he finds out we only go 100% on our machines at 3:17AM.I report the improved CPU and token utilization metrics during my performance review and negotiate a 300% raise.
>>108826670@grok
ask your admin to check whatever your monitoring tool says (zabbix or whatever). This is more of an IT question rather than backend coder.>inb4 no toolset installedthat's your first issue, probably the bigger problem here than small cpu spike
>>108826670Task Manager within the server where the CPU is wonky?
>>108826670I'm checking myself out, who cares if a computer does work at 3 am? It's probably fine.
The security camera tapes to ensure the rent-a-cop isn't doing something stupid at that time. You'd be surprised how often this type of thing has a physical cause due to something fucking around with something.
>>108826670>checkinglastlogs, new hires
>>108826670Make a tweet then screencap that tweet for a 4chan thread that looks like a joke, hoping someone will be tricked into helping me
>>108826670grep for '*3:17*' in /var/logs/* or check journalctlAlso this Indian didn't explain the context, but I'll assume a server running Linux
>>108826670>backendI check if any app has a daily garbage collection task, because obviously whatever retarded troglodyte sysadmin is talking to me about this already determined that my apps were causing the increased CPU usage.
>>108826670clearly I'm checking my twitter thread I just asked the question on because I'm trying to farm answers on a topic I don't know the solution for and stealing the correct one and presenting it as a fix
>>108826670>what are you checkingany open tickets about it? nope?if its not broke dont fix itgenuinely my company quality system needs trace all fixes to initial complaint. no complaint you cant fix it.
>>108826670>No x.>No y.>No z.ai slop detected, thread disregarded>sage: announced
>>108826670Infosec is probably triggering a heavy ass scan with one of the half dozen little security toys they required us to install.
>>108826670>Asks ChatGPT>The first command it tells me to use is "sar"
>>108826699it's called journal now grandpa, you're out
>>108826670I'm not applying for a senior position
htop
laraslop script injection
>>108828989there's atop that will log for you something ala htop results over time, htop/top is not it unless you plan to watch it at 3:17
>>108826699>>108828916Yeah you can't say logs black people used to sit on logs you racist pig.
>>108829030what about journal of uncle tom
>>108826670>run some logger>check the logs
>>108826670Based on the IT dept at my company, the answer is:>reboot the machine>hard reboot>critical server? doesn’t matter, reboot>claim it was an ‘unplanned outage’>claim problem is fixed (problem is not fixed)
>>108828877this
>>108828908>saarThat you already know the timing of the CPU spike implies sar or equivalent has already been used.
>>108828877yup, we need to re-ddos your server with 60k(rising every day) exploit checks to make sure you haven't downgraded to vulnerable log4j susceptible version fuck your cpu
>>108828083system logs aren't the only thing you check dickweed, it's just the first thing
The database
>>108826670>What are you checking first?Nothing. Who gives a shit about a bit of CPU usafe at 3:17AM?
>>108826670I don't need to check out anything. Either Surajesh Patel or Juan pedro Carlos runs a bitcoin miner on our system, just fire both to be safe.
>>108829019I would be up at 3:17, I don't go to bed till like 6am these days
>>108829130>saars as a resource
>>108826670www.worldtimebuddy.com, check the time in india, china, russia and israel, if any of those are close to dawn rangeblock them for 24h.
>>108826670Journalctl
>>108826670>3:17 AMGod wouldn't be up this late.
>>108826670thank god I no longer work in IT. they treat you like a janny.
Let me guess, it's an epic meme interview question where the "real" answer is "nothing, our systems are running at 100% CPU 24/7 because he have that grindset hustle culture 100x mindset"
>>108826670It's 100% some windows cuck update that is forcefully raping your pc despite you telling them no
>>108826688kek
>>10882955795% of a workday as IT is sitting on your ass doing nothing, and telling people to try to reboot their device when you fucked up the infra and everything is down for everyone. So yes, you do deserve to be treated like the jannies you are.
>>108826670>log on to server>open terminal>codex --yolo>type "explain to me in 1 sentence why every night at 3.17 AM CPU usage spikes to 100%. no guesswork, get me the exact reason"
>>108829563>Upgrade server CPU from 100% to 100%+AI>>108829578But there is no "traffic spike", which could be inbound or outbound, so presumably you aren't downloading anything.
>>108829745Best part is rebooting the device magically solves 95% of those issues that are actually local and not "down for everyone", jumping headfirst into serious investigation logs and traces and whatnot without a reboot is a typical midwit IT janny mistake.
>>108826670My coffee pot xD can't start a debugging sesh without my fix!!
>>108830238Case in point.>hurr durr y dey call jannies jannies durrCry about it
>>108826670dashboard and netstat
>>108826670I set up a scheduled task to log CPU usage data for 5 minutes, it starts running at 3:15 AM and stops at 3:20 AM. I make sure to announce this fact to all the employees. If the CPU usage stops going up after the announcement then I know it's just because some faggot is mining coins on the company servers.
>>108826747The only answer. But do it with any AI in general as well. Ask it to write a script that analyzes all processes that are running and log them. Have it filter out process and show only those that are >20% CPU usage. And then play a sound, when CPU hits the target threshold and log them properly. And have AI analyze what it is doing, what files its accessing, where the process location is, who the developers were and for what purpose, when it was installed, etc
>>108826747Grok, is this true ?
>>108826670First identify which process consumes CPU, then trace what scheduler or subsystem invoked it.t. I'm a backend seniort. that was a lie, I asked GPT
>>108827273used CPU is increased power usage which costs money
>>108826670My Bitcoin wallet
>>108826670the real answer is to check whether anyone cares with the managementno need to waste time on a problem that affects literally noonet. senior dev
>>108826670These two-word sentences are a huge redflag, I noticed only indians (and sometimes claude) type like this.>checks usernameCalled it again.
>what are you checking firstsad panda for the latest incest drop
>>108826780important detail;you begin doing this at 4 pm and report it as a security issue so you get paid for an entire day of shitposting and go to sleep at 3:16 AM because it's fucking windows messing with an update
>>108826670I’ll check if my Red Bull is in the refrigerator and drink it while L2 finds the culprit
>>108836446>senior back end>drinking red bullwell that's a bullet proof health for a back end senior
>>108826670Is it an application server? A DB server? Since it's for backend dev, I assume application server. I will also assume "no cron jobs" means no scheduled application jobs, e.g. Spring @Scheduled tasks. First step is application logs. If nothing suspicious turns up, I check firewall/ingress/load balancer logs depending on what the network architecture is, looking for suspicious requests, maybe ICMP attacks which wouldn't appear at the application level. Failing that, check OS level stuff. If it's a Windows server, check event viewer, check if it was doing some retarded update. If Linux server, check dmesg/syslog/journalctl. If nothing turns up, ignore it because a momentary spike at such an hour is unlikely to cause service disruption.
