[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: tune2fs: last write time is weeks or even months ago, moreover before last reboot

On 07/25/2014 10:31 AM, David Guyot wrote:
> Le vendredi 25 juillet 2014 à 00:27:30 +0200, Linux-Fan a écrit:
>> On 07/24/2014 03:44 PM, David Guyot wrote:
>> If it is a Hitachi, the raw read error rate can switch between 0, 1, and
>> the 2^16 without the disk being faulty. My only Hitachi disk also does
>> this and works quite well. I also remember seeing some blog/article on
>> the Internet explaining this behaviour but I do not currently find it.
>> Considering the rest of your issues, I do not know, but the slowness is
>> definitely a bad sign. Is a simple reboot an option?
> Thanks for your answer, Linux-Fan. In fact, hard drives are Toshiba, not
> Hitachi, but the problem may be the same.
> I considered a reboot with a forced all-FS check, but the slowness is so
> considerable that I'm afraid to stop hosted websites for hours if FS
> checks are as slow as standard I/O ; in addition, as our servers are
> rented and not physically accessible, we won't be able to monitor e2fsck
> to see its progress or potential problems. In fact, that's partly why I
> asked advices here: to avoid this potentially problematic full FS check
> as long as I can find other solutions.

I see. Then I'd try to monitor the slowness as good as possible,
possibly using something like `atop` or even uptime's loadavg display to
find out if these (especially atop) point to a specific component. Also
watch out for all sorts of file systems, especially those which are not
directly on HDDs: I have encountered strange slowdowns when a fuse
filesystem was failing or some network filesystem was not available...

> I'll try to find this article you quoted, maybe I will find some infos
> applying to my problem. Meanwhile, thanks for your help; if you have
> other details, I'm interested.

Getting more curious myself, I have tried to find it again, but the only
things I could find were two /other/ forum entries which describe
(probably) the same problem:

 * http://www.hartware.de/forum/viewtopic.php?f=16&t=23001

The second one is in German, it basically says:

	Raw Read Error Rate is fluctuating between 95 and 100 for
	"Current" and "Worst" (even randomly improves after

and later on

	Before: (Raw_Read_Error_Rate: 65536), After: changed from
	65536 to 1


	just rebooted and it is back to 65536

and the Hitachi support answer basically says the HDD is OK.

Unlike me, the OP of that forum thread decides not to trust the HDD (I
wonder why, mine has been running for 2800 hours without any problems now).


Attachment: signature.asc
Description: OpenPGP digital signature

Reply to: