| To: | Steve Lord <lord@xxxxxxx> |
|---|---|
| Subject: | Re: XFS for postgres databases? |
| From: | Olaf Frączyk <olaf@xxxxxxxxxxxxx> |
| Date: | Mon, 07 Jun 2004 10:16:33 +0200 |
| Cc: | Jeremy Jackson <jerj@xxxxxxxxxxxx>, Christoph Hellwig <hch@xxxxxxxxxxxxx>, Nathan Scott <nathans@xxxxxxx>, linux-xfs@xxxxxxxxxxx |
| In-reply-to: | <40C0D7A8.9050500@xxxxxxx> |
| References: | <200405121300.14866.stevew@xxxxxxxxxxxxxxx> <1084332608.11308.3.camel@noodles> <20040512143941.A389759@xxxxxxxxxxxxxxxxxxxxxxxx> <200405121618.27843.ncunningham@xxxxxxxxxxxxx> <20040512165624.C389759@xxxxxxxxxxxxxxxxxxxxxxxx> <20040512130047.A28089@xxxxxxxxxxxxx> <40C0D621.2010001@xxxxxxxxxxxx> <40C0D7A8.9050500@xxxxxxx> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
On Fri, 2004-06-04 at 22:12, Steve Lord wrote: > Note that the blockdev mapping and the xfs metadata are the same > memory, its just they do not cooperate on the locking correctly, > so if you access both at once, you end up trashing on or the other. > Hi, Any ETA when it is fixed? This is really dangerous. I think nearly nobody suspects that reading block device with mounted filesystem can produce any kind of corruption. I think that there should be some info in the kernel config, saying that it is not allowed to run ANY program which uses block device for reading (like hdparm) on device with mounted XFS filesystem. BTW., if I have XFS on /dev/sda1 is it unsafe to access /dev/sda1 only or the whole /dev/sda too? If the latter, it means that modyfying partition table is also an unsafe operation. This would be really ugly. Regards, Olaf |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: XFS deadlock (bounced, resend), Nathan Scott |
|---|---|
| Next by Date: | Re: XFS for postgres databases?, Christoph Hellwig |
| Previous by Thread: | Re: XFS for postgres databases?, Steve Lord |
| Next by Thread: | Re: XFS for postgres databases?, Christoph Hellwig |
| Indexes: | [Date] [Thread] [Top] [All Lists] |