1 |
Am 26.03.2013 15:13, schrieb J. Roeleveld: |
2 |
> On Thu, March 21, 2013 21:03, Grant Edwards wrote: |
3 |
>> On 2013-03-20, Carlos Hendson <skyclan@×××.net> wrote: |
4 |
>> |
5 |
>>> Could this be the cause of the stalls during compiles? If it is the |
6 |
>>> cause, is it possible for the kernel to detect such failures and report |
7 |
>>> them? |
8 |
>> I think that as long as the errors are "recovered" they shouldn't |
9 |
>> cause problems. The $64 question is whether those errors are |
10 |
>> indicating that you're headed for unrecoverable errors (which will |
11 |
>> cause problems). If they seem to be increasing in frequency, I'd |
12 |
>> probably be a little worried. |
13 |
> Consumer harddrives, which I am assuming you are using, will stall |
14 |
> indefinitely untill they have fixed the error they are encountering. |
15 |
|
16 |
no, they don't. |
17 |
PATa drives take 30secs. |
18 |
|
19 |
SATA not even a tenth. |
20 |
|
21 |
That is why dmesg can fill up with hdd errors and you don't realize it |
22 |
until a reboot and the drive not spinning up anymore. |
23 |
|
24 |
> |
25 |
> RAID-drives (With TLER = Time Limited Error Recovery) will return sooner |
26 |
> with an error-code which can then be handled by the controller and/or |
27 |
> driver. |
28 |
> |
29 |
> In other words, yes, these can explain the stalls during compiles. |
30 |
|
31 |
they would - if it wasn't already established that it was overheating. |