Skip to content

A library of things for turning log lines into sqlite rows

Notifications You must be signed in to change notification settings

frameable/log-parse

Repository files navigation

log-parse

log-parse is a library for parsing fluentd logs, and getting them into sqlite databases

Getting started

To add log-parse to your project, npm install --save @frameable/log-parse

Example code can be found examples

The ctx function

Reading logs

Given a directory with some logs, you can open a log iterator there. It will iterate all of the files in that directory, in order of most recently dated ( the date being determined by the filenanme - (I should write more docs for this )), and yield each log line. It will decompress gzipped files.

For example, consider the following directory that contains logs

$ pwd 
/var/log/my-app/

$ head current.log
2023-08-04T00:10:26+00:00 {"status": "OK"}
2023-08-04T00:10:27+00:00 {"status": "OK"}
2023-08-04T00:10:28+00:00 {"status": "OK"}
2023-08-04T00:10:29+00:00 {"status": "NOT OK", "message": "the server tripped and fell!"}

$ head file.20230803.log.gz | gunzip
2023-08-03T00:10:26+00:00 {"status": "OK"}
2023-08-03T00:10:27+00:00 {"status": "DEGRADED", "message": "the server is getting sleepy..."}
2023-08-03T00:10:28+00:00 {"status": "DEGRADED", "pending_events": 42}
2023-08-03T00:10:29+00:00 {"status": "ASLEEP", "message": "the server is conked out!"}

We can iterate those logs as they are. We're using the ctx function to specify a default context but with logfileRoot set to our target directory. The Context struct is going to be the interface to most of the API

for await (const log of iterLogs(ctx({logfileRoot: "/var/log/my-app"}))) {
  console.log(log.content) 
}

Parsing logs

If we want to parse out the JSON body, we can. The built-in generator entriesKV is perfect for this - it will regex match for a body. By default, it uses the expression /^(?<timestamp>[^\t ]+)[\t ](?<body>.+)$/, but any expression that matches a body and timestamp can be used with entryRegex

for await (const entry of entriesKV("my-app", iterLogs(ctx({logfileRoot: "/var/log/my-app"})), ctx())) {
  console.log(entry.body.status)
}

The function chunkEntries can be used to read and parse chunks of log lines at once, where those chunks are then yielded.

Putting logs in SQL

To actually get this stuff in the database we need to have a database. We can also make it on the disk at sqliteRoot

const database = makeDatabase("year-digits", ctx({sqliteInMemory: true}))

We can either use insert with the database and some data directly

const entries = const entries = await chunkEntries(entriesKV("year-digit", iterLogs(ctx({logfileRoot: "/var/log/my-app"})), ctx()), 4, 0).next() // the first chunk of 4
insert(entries.value, database, ctx({sqliteInMemory: true, entryFields: new Set("status", "message")}))

We can also create an insertFunc to call later on other collections of entries

const insFunc = insertFunc(database, ctx({sqliteInMemory: true, entryFields: new Set("status", "message")}))
for await (const chunk of chunkEntries(entriesKV("year-digit", iterLogs(ctx({logfileRoot: "/var/log/my-app"})), ctx()), 4, 0)) {
  insFunc(chunk)
}

entryFields describes what fields to create columns for. log-parse also creates meta fields:

  • identifier uniquely identifies a single log entry in the scope of all of the files in its logfileRoot. identifier is in chronological order
  • timestamp comes from the value captured by entryRegex, which is then parsed into a Date
  • data has a json blob with any entry kv pairs that weren't an entryField

Now we have a database that we can interface with regularly. By default our table is logs, that can be set with sqliteTable.

select * from logs;
-- 1691021426000|2023-08-03T00:10:26.000Z|{}|OK|
-- 1691021427001|2023-08-03T00:10:27.000Z|{}|DEGRADED|the server is getting sleepy...
-- 1691021428002|2023-08-03T00:10:28.000Z|{"pending_events":42}|DEGRADED|
-- 1691021429003|2023-08-03T00:10:29.000Z|{}|ASLEEP|the server is conked out!
-- 1691107826004|2023-08-04T00:10:26.000Z|{}|OK|
-- 1691107827005|2023-08-04T00:10:27.000Z|{}|OK|
-- 1691107828006|2023-08-04T00:10:28.000Z|{}|OK|
-- 1691107829007|2023-08-04T00:10:29.000Z|{}|NOT OK|the server tripped and fell!

select count(status), status from logs group by status;
-- 1|ASLEEP
-- 2|DEGRADED
-- 1|NOT OK
-- 4|OK

select json(data) from logs where data != '{}';
-- {"pending_events":42}

About

A library of things for turning log lines into sqlite rows

Resources

Stars

Watchers

Forks

Packages

No packages published