Unsolved
This post is more than 5 years old
2 Posts
0
23445
June 27th, 2012 10:00
R410 running Linux: sometimes fails to detect SATA drive
I have encountered a slightly worrying condition whereby R410s sometimes fail to detect an attached SATA HDD, or sometimes the CD-ROM. It's random: power off, and on again, or even just reboot, and the "missing" drive will probably be visible again. The drive which "disappears" is also random - it may be the first physical drive, the second physical drive, or the CD-ROM. This has been observed on R410s, with BIOS versions from 1.5.x to 1.8.2. Since we use software RAID (Linux "md"), we're spending a lot of time re-mirroring "broken" arrays especially on test systems. I don't think any of the servers this has happened on have even got a hardware RAID controller physically installed (or we're using the simplest controller, in "hba" mode).
This has been observed booting "syslinux" (for installation); booting an installed 2.6.32 kernel (Debian Lenny); and booting an installed 2.6.39.4 kernel. The symptom normally is that "md" becomes degraded, and only "/dev/sda" is present. Tracing through dmesg will show that "/dev/sda" may be either SCSI 0:0:0:0, or 1:0:0:0 (it may be either one of the physical drives). What's worse, following a reboot I have observed the reverse: the previously missing drive is now present, as /dev/sda, and the previously present drive is now missing. With software RAID, this results in apparent "time travel" - as the system reverts to a now-outdated mirror, not even knowing it's out-dated!
I have not been able to find any references to this situation, either generally through searches, or through the Dell Linux forums. Has anyone else observed this, and does anyone have a clue what's going on?
DELL-Jonathan S
153 Posts
0
June 27th, 2012 11:00
I am at a loss to explain the symptoms you're observing, but I will point out we have released BIOS 1.10.2 just yesterday (26 June) as well as 1.9.0 in December. 1.10.2 is available from www.dell.com/.../DriverDetails
Since you are seeing trouble with the CD-ROM as well I would tend to look towards the SATA controller and related components rather than the hard drives specifically. How frequently does a given box typically exhibit the symptoms you have described? Are all SATA devices recognized at least until booting is complete, or have you sometimes seen a drive missing/offline as early as POST? Might you be able to test one of these with RHEL/CentOS 6.x which also runs 2.6.32? I just suggest that since we have fully support RHEL and while the kernel version from Debian is the same they have different patches and configuration options and if they behave differently (or the same!) it may help isolating the issue.
ajsealy
2 Posts
0
June 28th, 2012 07:00
Thank-you, I will see about getting our BIOSs updated to current. I honestly can't say if the drives showed as missing during POST, since most of these servers are "headless", but I can ask for this to be watched out for. I do recall at least one instance where no CD-ROM was reported during POST, although it was present, and did re-appear after power cycling. All cables were snug, I checked that.
I do know that, when they go missing, this is very early in the boot process, and happens with both 2.6.32 and a much more recent, fully patched 2.6.39. If BIOS doesn't fix the problem, or it does appear that the drives are detected during POST then subsequently not being detected, I can try a CentOS 6.x install.
r410ata
1 Message
0
April 4th, 2013 17:00
I would like to confirm I am experiencing the same issue on Centos with 2.6.35
A power cycle is the only solution.