Which WIFI USB dongle works best with Centos5?

[sbeh]

Hi there,

could anyone here report from experience which WIFI USB dongle "just works" with Centos 5.2?

Thanks

/Stefan.

[pschaff]

I think you meant to say "CentOS 5.3". :-)

A forum search for "usb wifi" shows some past recommendations:
http://www.centos.org/modules/newbb/viewtopic.php?viewmode=flat&topic_id=18941&forum=40

It will also turn up some to stay away from.

I'd be interested in more just-works recommendations as well.

[sbeh]

I ran exactly that search before posting but it did not seem to produce any recommendations for concrete products that have proven to work.

And, yes, I meant Centos 5.3.

It seems that Ralink chipsets are a general recommendation given in other places.

Then the problem is "just" to find a product that you can be sure has exactly the chipset that is supported.


Thanks

br/Stefan.

[sbeh]

Ok, let's assume there is no USB WIFI dongle that "just works" with Linux. ;)

I would like to summarize my experience with the Edimax EW-7318USG for future reference.

The good things:
- Has a detachable antenna, so if you ever need more gain, you can fix it
- Centos 5.3 recognizes it and uses the rt73usb driver (no build needed)
- WPA2/PSK works with the Centos-provided wpa_supplicant (-Dwext)
- Stable connection, at least for me. But note that radio connections depend on various factors so your results may vary.
- Supports monitoring mode with Kismet (source=rt2500,wlan0,RT73)

In order to make it work, I had to do two things:
- Install the firmware package. It's not in the Centos repos, although RH supports it (http://rhn.redhat.com/errata/RHEA-2009-0061.html). But it can easily be found via Google.
- Revert back to Centos kernel 2.6.18-128.1.1 to get rid of a kernel panic during reboot. This seems to be a general kernel problem that affects various drivers and adapters. Let's hope it's only temporary...

Summary; works well, but not without hassle.

br/Stefan.

[AlanBartlett]

Thank you for that valuable report, [b]Stefan.[/b]

You have covered a number of points which, at present, seemed a little "grey".

[sbeh]

I have to correct myself slightly.

- The RX/TX counters as shown by ifconfig -a don't count.
- Kismet only works sometimes. Sometimes it quits after a couple of seconds with a msg saying that the server sent EOF.
- It happened twice within the last 3 days that I returned to my computer and found the 2 LEDs on the kbd blinking, i.e. the kernel had crashed.

The latter has never happened before on that machine, with any OS.

I doubt that above problems are specific to the rt73usb driver. But who knows.

Linux sucks, really.

/S.

[pschaff]

[quote]
sbeh wrote:
...
I doubt that above problems are specific to the rt73usb driver. But who knows.
[/quote]
Stefan,

I would strongly suspect the driver, as driver issues like this are usually the cause of relatively rare Linux crashes.

[quote]
Linux sucks, really.
[/quote]
That seems a rather broad and unsupported conclusion. With Linux at least the source of the driver is available and many people are willing to work to fix problems. Sorry you are having problems, but a more constructive response would be to file a bug report.

Don't think you have revealed details of the system, nor the issue that has kept you from using the latest kernel, which usually fixes the most issues. Is it the DELL D630 with Intel 3945 you posted about elsewhere (and contributed to a bug report on)?

[sbeh]

[quote][quote]
Linux sucks, really.
[/quote]
That seems a rather broad and unsupported conclusion. With Linux at least the source of the driver is available and many people are willing to work to fix problems. Sorry you are having problems, but a more constructive response would be to file a bug report.

Don't think you have revealed details of the system, nor the issue that has kept you from using the latest kernel, which usually fixes the most issues. Is it the DELL D630 with Intel 3945 you posted about elsewhere (and contributed to a bug report on)?
[/quote]
That was a generalization, of course. ;)
More specifically though, it sucks big deal on the desktop.
Since Redhat 7.3, across many Fedora releases up until Centos 5.3 it has cost me hundreds of hours to fix things like DMA harddisk access, USB mass storage, gfx acceleration, fonts, sound, video playback, and WIFI.

It is discouraging to see that even on the presumably most mature Linux distro simple things like USB mass storage still don't work reliably with anything but the most trivial setup (try plugging in/out more than one USB storage device with the KDE version that ships with Centos 5.3, maybe use harddisks with several partitions, to make it worse)

Quite honestly, I have come to believe that Linux is an experimental operating system for geeks who find more fun in fixing problems rather than using the computer to do something useful (again I'm reflecting my subjective impression using it on the desktop for about 6 years).

[/rant]

I'm now having fun with an EDIMAX EW-7318USG USB WIFI (Ralink/rt73usb) on a desktop machine. When I first reported kernel crashes in the rt73 developers forum, they suggested it may be the mac80211 framework. The Centos bug report you refer to describes similar problems on Intel hardware, so it could be they're right, and it's not the driver.

The most recent Centos kernel panics *every time* I stop wpa_supplicant, plug out the WIFI, or simply reboot.
Based on comments elsewhere I reverted back to -128.1.1 which allows me to reboot cleanly, but it still shows these occasional kernel crashes.

BTW what is the status of mac80211 that's now in the kernel? Do people call it "stable" or is it officially still "experimental"?

br/Stefan.

[pschaff]

[quote]
sbeh wrote:
...
That was a generalization, of course. ;)
More specifically though, it sucks big deal on the desktop.
Since Redhat 7.3, across many Fedora releases up until Centos 5.3 it has cost me hundreds of hours to fix things like DMA harddisk access, USB mass storage, gfx acceleration, fonts, sound, video playback, and WIFI.
[/quote]

RHEL-derived distros have never been the strongest for the desktop, OTOH, after using Fedora, Ubuntu, and a few outliers, I keep coming back to them.
AS far as the hours involved, that is likely to be true of [b]any[/b] OS. I'd hate to see an accounting of the hours I've spent on Windows issues over the years.

[quote]
It is discouraging to see that even on the presumably most mature Linux distro simple things like USB mass storage still don't work reliably with anything but the most trivial setup (try plugging in/out more than one USB storage device with the KDE version that ships with Centos 5.3, maybe use harddisks with several partitions, to make it worse)

Quite honestly, I have come to believe that Linux is an experimental operating system for geeks who find more fun in fixing problems rather than using the computer to do something useful (again I'm reflecting my subjective impression using it on the desktop for about 6 years).

[/rant]
[/quote]
Quite honestly I disagree. Sounds like typical sysadmin issues on any OS, but then I'm a geek who has gone through Burroughs B5500, early IBM timeshare, CDC mainframes, Commodore Vic20/C64, MS/PC-DOS, Amiga, WfWG-3/3.1/3/11 - XP, VAX/VMS, SunOS, Solaris, etc., etc., etc. so take it with the requisite grain of salt. :-)
...[/quote]

[AlanBartlett]

[quote]
I'm now having fun with an EDIMAX EW-7318USG USB WIFI (Ralink/rt73usb) on a desktop machine. When I first reported kernel crashes in the rt73 developers forum, they suggested it may be the mac80211 framework. The Centos bug report you refer to describes similar problems on Intel hardware, so it could be they're right, and it's not the driver.

The most recent Centos kernel panics *every time* I stop wpa_supplicant, plug out the WIFI, or simply reboot.
Based on comments elsewhere I reverted back to -128.1.1 which allows me to reboot cleanly, but it still shows these occasional kernel crashes.
[/quote]

Please see [url=http://www.centos.org/modules/newbb/viewtopic.php?topic_id=20384&forum=40&post_id=78431#threadbottom]this forum thread[/url].