Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

https://github.com/stewartsmith/libeatmydata

libeatmydata - because fsync() should be a no-op
https://github.com/stewartsmith/libeatmydata

Last synced: 6 days ago
JSON representation

libeatmydata - because fsync() should be a no-op

Host: GitHub
URL: https://github.com/stewartsmith/libeatmydata
Owner: stewartsmith
License: gpl-3.0
Created: 2013-09-26T06:54:47.000Z (almost 11 years ago)
Default Branch: master
Last Pushed: 2024-01-07T17:44:16.000Z (6 months ago)
Last Synced: 2024-03-06T17:39:14.664Z (4 months ago)
Language: C
Homepage: https://www.flamingspork.com/projects/libeatmydata/
Size: 593 KB
Stars: 371
Watchers: 16
Forks: 20
Open Issues: 12
Metadata Files:
- Readme: README.md
- License: COPYING

Lists

awesome-ld-preload - libeatmydata - disable all forms of writing data safely to disk. fsync() becomes a no-op, O\_SYNC is removed, etc. (Libraries)
awesome-stars - stewartsmith/libeatmydata - libeatmydata - because fsync() should be a no-op (C)
awesome-stars - stewartsmith/libeatmydata - `★395` libeatmydata - because fsync() should be a no-op (C)

README

libeatmydata
============

An LD_PRELOAD library that disables all forms of writing data safely to disk.
fsync() becomes a NO-OP, O_SYNC is removed etc.

The idea is to use in testing to get faster test runs where real durability is
not required.

***DO NOT*** use libeatmydata on software where you care about what it stores. It's called lib***EAT-MY-DATA*** for a *reason*.

see http://www.flamingspork.com/projects/libeatmydata

Installing
------------

Debian/Ubuntu and friends:
```
apt install eatmydata
```

To build from source, you will need `autoconf`, `autoconf-archive`, `automake`, `libtool`, and a compiler (say, `gcc`).

Then, build and install like so:
```
git clone https://github.com/stewartsmith/libeatmydata.git # (or extract from source tarball)
cd libeatmydata
autoreconf -i # (if building from git only)
./configure
make
make check
sudo make install
```

Usage
-----

```
eatmydata foo
```

Performance Improvements
------------------------

When running part of the MySQL test suite in 2007 on my laptop:

```
TEST RESULT TIME (ms) TIME (with libeatmydata)
----------------------------------------------------------------------------------

main.innodb-lock [ pass ] 4523 4323
main.innodb-replace [ pass ] 102 56
main.innodb-ucs2 [ pass ] 5786 1084
main.innodb [ pass ] 78306 24900
main.innodb_gis [ pass ] 2647 1544
main.innodb_mysql [ pass ] 86810 68579
main.innodb_notembedded [ pass ] 198 150
main.innodb_timeout_rollback [ pass ] 2990 2750
main.innodb_trx_weight [ pass ] 1608 841
---------------------------------------------------------------
Stopping All Servers
All 9 tests were successful.
The servers were restarted 7 times

WITHOUT: Spent 182.97 seconds actually executing testcases
WITH : Spent 104.227 seconds actually executing testcases

WITHOUT: WITH:
real 3m36.053s real 2m10.610s
user 0m42.323s user 0m41.939s
sys 0m2.844s sys 0m2.356s
```

Talks/Video
-----------

libeatmydata was the product of a talk I gave back at linux.conf.au 2007, titled
"Eat My Data: How Everybody gets File IO Wrong" - I also gave this talk at OSCON.
This talk went over the common mistakes people make when using POSIX file IO
routines.

You can watch a video recording of this talk at:

- https://youtu.be/LMe7hf2G1po
- [http://www.linux.org.au/conf/2007/talk/278.html](https://web.archive.org/web/20130504133047/http://www.linux.org.au/conf/2007/talk/278.html)