Why We Recommend Against OpenFiler
-
Original thread at Why We Recommend Against OpenFiler on SpiceWorks. The info is so important, it should be here, too. Originally from August, 2013. The current release version is still 2.99. So four years they've not even updated the website except for the copyright at the bottom.
After a few years [now nearly eight] of recommending against the use of OF for storage projects, I'm surprised to find that it still comes up in nearly daily conversations as a tool being considered for storage. The reasons that we recommend avoiding OpenFiler are pretty concrete issues that should rule out any product for use in storage. Even in a lab, OF does not make sense.
Here are the issues, roughly in order of criticality:
- Users of OF report, even when all things are done correctly, that data will randomly vanish on reboots. The entire storage subsystem drops and the data is simply gone.
- The included iSCSI (SAN) stack is one that lacks SCSI Reservations making it unsupported when used with multiple connections (the only way that SAN is useful.) Things that are particularly affected by this would be virtual machines or databases - exactly where you think you would use a SAN for safety is specifically where your data corrupts. The very reason to have installed a SAN is the very thing OF doesn't do reliably. If you use OF as a SAN the expectation is that even if it doesn't lose your data it will eventually corrupt it. OF sells a replacement SAN / iSCSI product for a very high price for OF so they recognize the issue but act as if corruption of your data is acceptable and that this is a "performance" issue when it is actually a reliability issue.
- OF is built on the abandoned (100% discountinued, the company is gone) niche rPath distribution. rPath was a bad idea when it was still actively maintained. Today using rPath is outright crazy. This means that the patching and updating mechanism on which OF relies is gone. This also means that not even Linux experts are familiar with the quirks of this niche platform. It uses an abandoned packing system called Conary that is universally loathed and never got on. So OF is, in some ways, abandoned, at least in its current form, by extension.
- Clustering isn't crippled but is not included and even if set up manually (an extremely expert level task) the OF interface does not support it and it breaks the entire concept of OF in the first place.
- OF's support community is practically inactive and there is more OF activity in Spiceworks than in the OF forums. If you want support, you'll need to pay for it from the same guy that made the bad decisions mentioned above.
- OF appears to be defunct. Some Spiceworks community members were suspicious about four years ago that the company was gone and only the website was left so we called in. We believe we reached someone's wife who attempted to persuade us that the company did actually exist but was unable to produce any employees. We think they are using a home number. At that point, no OF had been released in so long we thought the project was dead. Four years later the only "release" is 2.99 which is numerically a beta release from several years ago. So no final release has come out in many, many years and even 2.99 has sat static for several now. If it wasn't an abandoned project in 2010, it almost certainly is now.
- If you do pay for support it appears that there is only one employee. This suggests that support is a high risk scenario - if he is not available or not able to fix the issue, that's it.
- The web interface is confusing and basically pointless. A normal install of Linux, without the web interface, is arguably easier to use for the same tasks. Using OF makes all tasks harder and adds risk for no reason.
- OF is so old that it lacks many important storage features that are broadly available on other Linux distros today such as updated DRBD, EXT4, BtrFS, ZFS, etc.
The bottom line is that OF essentially offers no advantage over just using OpenSuse, CentOS, Ubuntu, Debian, Mint, etc. but has many caveats, some minor and some staggeringly huge (like your data will just vanish or corrupt.) Stability is not something to toy with when it comes to storage.
-
And on top of all of that, it is subject to the Jurassic Park Effect, as well.
-
-
@scottalanmiller Just found this old thread from 2012 that is pretty great, as well:
https://community.spiceworks.com/topic/281007-challenging-openfiler
-
@scottalanmiller said in Why We Recommend Against OpenFiler:
Originally from August, 2013. The current release version is still 2.99. So four years they've not even updated the website except for the copyright at the bottom.
Following up a few more years later. No further release has ever come. 2.99 remains the only available download. The only change is that in December, 2018, they posted on the download site that the download isn't patched for a major vulnerability because their upstream OS was abandoned in 2014. So at least they've recognized it as an abandoned project at this point.
-
You know, it is weird, every few years I check back because I have this weird fear in the back of my mind that maybe I was wrong in ~2011 when I researched OpenFiler so much and called them directly even to try to find out what was going on, looking into their broken network stack, data loss with clients, lack of updates, proposed fixes, made an open challenge for good community interaction. Calling out a vendor always worries me. But it is funny, it's been six years now since we posted why it was dead and eight years at least since calling them out for bad design to force users to pay for support to stop losing data (data extortion, as it is known.) In that time, no fixes, no patches, no replacement product, no apologies. We got threats and nothing else. No idea why I still worry that nearly a decade later maybe they'll make a viable product and "prove something", lol.
-
my god openfiler is still around? I remember playing with it and wanting to use it but never ending up using in the end. One of the reason was it barely updated.
-
@360col said in Why We Recommend Against OpenFiler:
my god openfiler is still around? I remember playing with it and wanting to use it but never ending up using in the end. One of the reason was it barely updated.
Not really, apparently died a LONG time ago.
-
@scottalanmiller said in Why We Recommend Against OpenFiler:
Original thread at Why We Recommend Against OpenFiler on SpiceWorks. The info is so important, it should be here, too. Originally from August, 2013. The current release version is still 2.99. So four years they've not even updated the website except for the copyright at the bottom.
Just a follow up as I had totally forgotten about this product until I saw it mentioned in another ancient threat.... if you visit their page now, ANOTHER four years after this thread was originally made, OF is STILL on version 2.99. So eight years now, almost nine, since an update after their swore up and down that they were alive and doing well and all kinds of new stuff was coming out. Their only update was in 2018 to note on the website that the product was legacy and the kernel wasn't stable.
Eight years, no patches.
-
@scottalanmiller said in Why We Recommend Against OpenFiler:
You know, it is weird, every few years I check back because I have this weird fear in the back of my mind that maybe I was wrong in ~2011 when I researched OpenFiler so much and called them directly
A full decade now. Man that seemed way more recent.
-
@scottalanmiller said in Why We Recommend Against OpenFiler:
I saw it mentioned in another ancient threat.
freudian slip?
-
@danp said in Why We Recommend Against OpenFiler:
@scottalanmiller said in Why We Recommend Against OpenFiler:
I saw it mentioned in another ancient threat.
freudian slip?
LOL, indeed.