LTTNG-SCHEDLOG-MI(1)

💖 NAME 💖

okay so bestie, meet lttng-schedlog-mi — the LTTng analyses Scheduler log machine interface. that's the whole vibe. it's a user command, it lives in section 1 of the manual (you know, the LTTNG-SCHEDLOG-MI(1) one 💅), and it is here to make your scheduler logs make sense. periodt.

✨ DESCRIPTION ✨

here's the whole thing, exactly how you'd type it:

usage: lttng-schedlog-mi [-h] [-r REFRESH] [--gmt] [--skip-validation]
       [--begin BEGIN] [--end END] [--period PERIOD]
       [--period-captures PERIOD_CAPTURES] [--period-begin PERIOD_BEGIN]
       [--period-end PERIOD_END] [--period-begin-key PERIOD_BEGIN_KEY]
       [--period-end-key PERIOD_END_KEY] [--period-key-value
       PERIOD_KEY_VALUE] [--cpu CPU] [--timerange TIMERANGE]
       [--progress-use-size] [--no-intersection] [-V] [--debug]
       [--no-color] [--no-progress] [--min MIN] [--max MAX] [--procname
       PROCNAME] [--tid TID] [--freq] [--freq-resolution
       FREQ_RESOLUTION] [--freq-uniform] [--freq-series] [--limit LIMIT]
       [--top] [--log] [--stats] [--total] [--per-tid] [--per-prio]
       <path/to/trace>

The sched command. that's it, that's the whole summary. iconic.

🎀 POSITIONAL ARGUMENTS 🎀

• <path/to/trace> — the path to your trace. like, where the tea actually lives. give it the trace path and she'll do the rest 🫶

💅 OPTIONAL ARGUMENTS 💅

okay this is the long part bestie, but every single one of these matters, no skipping:

• -h, --help — shows this help message and then dips. bye 👋
• -r REFRESH, --refresh REFRESH — the refresh period, and you can tack a units suffix on it if you want (default unit: s).
• --gmt — manipulate timestamps based on GMT instead of local time. timezone shift, but make it intentional 🌍
• --skip-validation — skip the trace validation. living dangerously, we respect it.
• --begin BEGIN — start time, formatted as hh:mm:ss[.nnnnnnnnn].
• --end END — end time, same format: hh:mm:ss[.nnnnnnnnn].
• --period PERIOD — Period definition.
• --period-captures PERIOD_CAPTURES — Period captures definition.
• --period-begin PERIOD_BEGIN — the analysis period start marker event name.
• --period-end PERIOD_END — the analysis period end marker event name (heads up: this one requires --period-begin, they're a package deal 💞).
• --period-begin-key PERIOD_BEGIN_KEY — optional, a list of event field names used to match period markers (default: cpu_id).
• --period-end-key PERIOD_END_KEY — optional, a list of event field names used to match period markers. if you don't specify any, it just uses the same as --period-begin-key.
• --period-key-value PERIOD_KEY_VALUE — optional, define a fixed key value that a period has to correspond to in order to even be considered. picky, but in a slay way.
• --cpu CPU — filter the results down to only this list of CPU IDs.
• --timerange TIMERANGE — time range, written as [begin,end].
• --progress-use-size — use trace size to approximate progress.
• --no-intersection — disable stream intersection mode.
• -V, --version — show the program's version number and exit. quick lil flex.
• --debug — enable debug mode (or, alternatively, set the LTTNG_ANALYSES_DEBUG environment variable).
• --no-color — disable colored output. (sad, but sometimes you gotta.)
• --no-progress — don't display the progress bar.
• --min MIN — filter out durations shorter than min usec.
• --max MAX — filter out durations longer than max usec.
• --procname PROCNAME — filter the results only for this list of process names.
• --tid TID — filter the results only for this list of TIDs.
• --freq — output the frequency distribution of sched switch latencies.
• --freq-resolution FREQ_RESOLUTION — frequency distribution resolution (default 20).
• --freq-uniform — use a uniform resolution across distributions.
• --freq-series — consolidate the frequency distribution histogram into a single one.
• --limit LIMIT — limit to top X (default = 10).
• --top — output the top sched switch latencies. main character behaviour honestly.
• --log — output the sched switches in chronological order.
• --stats — output sched switch statistics.
• --total — group all results (applies to stats and freq).
• --per-tid — group results per-TID (applies to stats and freq) (this is the default, so it's already doing this unless you say otherwise).
• --per-prio — group results per-prio (applies to stats and freq).

💀 reminder bestie: --period-end will not work without --period-begin. don't come for me when it yells at you.

that's the whole page, no cap. this was LTTng Analyses v0.6.0, dated January 2017 — go forth and read your scheduler logs like the icon you are 💖✨👑