Home opinion-and-analysis Open Sauce Journald: just a reinvention of the wheel?

Two months back, Red Hat developers Lennart Poettering and Kay Sievers proposed a fundamental change to the structure of Linux in the shape of the Journal daemon, meant to replace the venerable syslog.

Syslog, as the name implies, provides a means of logging activity in a system. It has been around for about 30 years; an improved version known as rsyslog has been created by developer Rainer Gerhards, and is used as the default on Debian systems.

There has been much chatter about the virtues and evils of replacing syslog with Journald; Poettering, it must be remembered, is the man behind systemd, which serves as a replacement for the Sys V init scripts on some GNU/Linux distributions, and also PulseAudio, which, as the name implies, is a replacement for the sound management applications on GNU/Linux systems. His projects generally do invite comment.

Some developers have even started a petition online against Journald.

Journald is tied to systemd; this means Debian will not use it whenever it does emerge, the reason being that Debian has a port that uses the FreeBSD kernel. Systemd can be used only with the Linux kernel. A test version of Journald has just been released along with systemd v38 but a smaller number of bigger features are lacking.

Poettering has advanced in detail the reasoning behind the proposal for Journald and many of his reasons seem to be logical. But many developers are wary of one aspect of his proposal - the fact that the journal file format will not be standardised and will be varied as the creators see fit.

Senior Linux kernel developer Greg Kroah-Hartman is supportive of the idea and does not see it as reinventing the wheel. He also does not see it as a Red Hat-ism and thinks that there is plenty of logic to the claims made by Poettering and Sievers about the superiority of what they propose, over the existing implementations.

And Kroah-Hartman does not read too much into the statements that the file format will not be standardised. "Since when is it going to be undocumented?" he asked in response to queries by iTWire. "The developers are just stating that at the moment, as the project is young and not fully finished, they aren't going to guarantee that the log format is stable, and as such, it's 'undocumented'."

He said he did not see what the fuss was about. "I really do not know, (maybe) people don't like change?" he said, adding, "given that there is not even any code that has been produced yet, and given that the track record of the developers creating this code is _very_ good, I am not really worried at all about this."

A good part of the opposition to the Journald proposal has coalesced around the Linux Weekly News website; an article about the proposal by editor Jonathan Corbet in November, in which he appeared to lean at least slightly over to the Poettering-Sievers proposal, drew more than 200 comments, most of them supporting the status quo.

In summation, Corbet wrote at the time: "Unless a shinier SunOS 4.x is really all that we are after, we as a community are desperately in need of people who are willing to stir things up and try new things to solve problems.

"To restrict ourselves to what Unix did is an almost certain way to share Unix's fate. We can do better than that - indeed, we have done better than that - but, to do so, we need to create an environment where developers are willing to take some technical risks.

"Bad ideas need to be exposed as such whenever possible, but ideas should not be shot down just because they do not preserve the way our grandparents did things. The Journal may or may not turn into a sensible option, but the process that creates things like the Journal is one of the most important things we have."

Corbet told iTWire that his views had not changed since he wrote the article. "I've taken no position on whether the journal is the right solution, but I do agree that Linux as a whole does logging and messaging poorly, and I think it's good that people are willing to try to shake things up in this area," was his response.

Asked whether Journald might be another Red Hat replacement which could provide it with a competitive edge over other Linux distributions, Corbet replied: "That seems like a rather un-objective statement for a proper journalist. Care to name the other such replacements that this is presumably a part of? Bear in mind that, due to the conservative nature of RHEL, it is quite common that Red Hat's contributions to the community are first shipped by other vendors - doesn't seem like much of a competitive edge, somehow."

Last month, well-known journalist Brian Proffitt wrote a piece headlined "Red Hat's Linux changes: Fixes or ISV positioning? in which he questioned the rationale behind the Poettering-Sievers proposal.

Added Corbet: "One could say that the journal is an attempt to make a far-future version of RHEL more appealing in enterprise environments, where the shortcomings of current Linux logging are well known. But it's free software; anybody can ship it.

"Anybody can contribute to it and influence its direction, and there is no copyright assignment requirement, so there are limits to Red Hat's control over it."

Veteran Debian GNU/Linux developer and SE-Linux expert Russell Coker said that while the creation of Journald was reinventing the wheel, "sometimes things need to be rewritten. The list of features sounds quite interesting. The use of binary files does have a downside in that it will be more difficult to extract data when things go wrong. The UNIX tradition has been to have simple modular things connected with shell scripts and with plain text files used for most things, this tradition has worked well.

"But sometimes there are benefits that are compelling enough to make it worth changing."

Russell referred to a blog post of his own, saying, "Systemd has some great features that will improve system boot times, I've briefly commented about it. While systemd is more complex and more difficult to get working it has real benefits that make it worth using.

"After replacing init the next logical step is to consider what other essential parts of the boot process can be improved and syslog is a good candidate."

Asked whether Red Hat was trying to gain a competitive edge, he replied: "I don't think that Red Hat have any plans to gain a competitive edge over other Linux distributions in such things, that's not how they work. Also, it is free software, so it's quite possible for some other distributions to include it before it's included in a RHEL release, and given the amount of time taken to prepare and test a RHEL release it should be possible to push a release of a smaller distribution in that time period."

Russell said he had tried out systemd on some of his test systems and planned to migrate all his systems to it as soon as convenient. "I will try out the new logging too as it seems to provide some useful features," he added.

A full version of Journald is expected be released with version 17 of Red Hat's community distribution, Fedora; given the release schedule that the distribution has followed so far, it should be available to the world at large by the second week of May.


Download an in-depth guide to managing a healthy, motivated and energetic workforce without breaking the bank.


Sam Varghese

website statistics

A professional journalist with decades of experience, Sam for nine years used DOS and then Windows, which led him to start experimenting with GNU/Linux in 1998. Since then he has written widely about the use of both free and open source software, and the people behind the code. His personal blog is titled Irregular Expression.






Join the iTWire Community and be part of the latest news, invites to exclusive events, whitepapers and educational materials and oppertunities.
Why do I want to receive this daily update?
  • The latest features from iTWire
  • Free whitepaper downloads
  • Industry opportunities