disk crashes provoked by tahoe-lafs?

Greg Troxel gdt at ir.bbn.com
Tue Apr 15 14:47:15 UTC 2014


Daira Hopwood <daira at jacaranda.org> writes:

> On 15/04/14 01:34, Greg Troxel wrote:
>> 
>> I run tahoe servers on 4 systems in a private grid.  The grid is not
>> used much, but I run deep-check --repair --add-lease every week or so,
>> when I remember.  The nodes all have lease expiration turned on, but are
>> quite unfull.  All are running NetBSD, some -5, some -6.  I do not have
>> these filesystems mounted noatime.
>> 
>> 3 of these nodes are physical machines, with 3 kinds of disks.  All have
>> turned up bad blocks.  The 4th is a Xen domU; the dom0 is RAID-1.  One
>> of the dom0's disks also had some bad blocks.  This is a notable
>> cluster; disk failures are otherwise relatively rare.
>> 
>> So I really wonder if the lease checking code, or something, is churning
>> the disk.
>> 
>> Has anyone else seen this?
>
> 'tahoe deep-check --repair --add-lease' will write to the header of
> every share.

Thanks.  I really don't do the --add-lease all that often.  I had
wondered if there was some process that would be causing writes
e.g. every 30s on an otherwise idle system, perhaps to track lease
expiration.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 180 bytes
Desc: not available
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20140415/09579f38/attachment.pgp>


More information about the tahoe-dev mailing list