error issuing READ_DMA command

V

vuce

Dear fellow freebsd-ers!

Since update to freebsd 8.1 I started to get errors like this:

Code: 
unknown: TIMEOUT - READ_DMA retrying (1 retry left) LBA=0
ata2: timeout waiting to issue command
ata2: error issuing READ_DMA command
...
Sometimes it works ok for a while, but eventually the system crashes - there is a noticeable slowdown as zfs desperately tries to correct the errors - seems like it's always one of 3 drives in the main zfs pool.

I've ran wd tester which didn't detect any hard drive errors.

I've googled about this error but didn't find anything (useful) other than that it might be an intel controller "misunderstanding" with freebsd.

The mb is msi G41M-P33 (http://www.msi.com/index.php?func=downloaddetail&type=manual&maincat_no=1&prod_no=1914)

Thanks for any help in advance,
Miha
 
check the disk with:
[CMD=""]diskinfo -ctv <hdd>[/CMD]
any weird speeds ?

and it's Attributes with:
[CMD=""]smartctl -A <hdd>[/CMD]

look for 7 Seek_Error_Rate and 199 UDMA_CRC_Error_Count. Although SMART can be misleading, it can give you some idea about what's going on with your hdd.

In the meantime, maybe it would be a good idea to start backing up your disk.

One more thing ... is the hdd connected to a separate controller ?
 
@da1
Nothing on both counts. I really don't think it's HDD issue, since wd tester didn't find any errors on any of the drives.

"seems like it's always one of 3 drives in the main zfs pool." might have been a bit ill-defined: there are 3 drives in zfs pool, and the error links to one of the three, but it's not always the same one.

All hdds are connected to the sata controllers on the mb.

@olav
i was advised to fresh install the system on another drive but as I don't have any extra drives to spare I'd like to see if this might be solved in another way. Ultimately I will just buy another drive for the system.
 
I'm just guessing here but since you use ZFS and you did an upgrade, the pools themselves are using the old version of ZFS. 8.1 has v14 and 8.0 has v13 (AFAIK).

I'm not ZFS expert/user but I've seen that this can cause some issues.
 
@da1
I've updated the pool to v14 after the freebsd upgrade.

@User23
I've had the same idea, but G41M-P33 apparently doesn't support achi - at least there's no setting in bios and no mention of it in the manual.
 
So, if there are no more ideas I'll try opensolaris and see what happens. If it works than it's obviously some freebsd bug?
 
it would be a bug if you (or more ppl) can reproduce it.

try reinstalling everything and then do an upgrade to 8.1. if the behavior is the same, it could be a bug.

PS: maybe you should write the steps you took upto the point where you start receiving the error messages
ex: installed 8.0 with make.conf, etc,etc
upgrade to 8.1 using method X, etc....


again, try to reproduce it
 
Tried importing the pool to solaris but it says it has corrupted data. Zpool status on freebsd says everything's ok. Go figure.
 
Seems like a pretty usual problem, since solaris thinks adx is a slice instead of a whole hd... still searching for a workaround though.
 
An update: installed opensolaris, within a day same errors appeared - I'm assuming this has to be a hardware issue.
 
You must log in or register to reply here.
Back
Top