| 1 |
On Thursday 24 March 2011 15:38:02 J. Roeleveld wrote: |
| 2 |
> On Thu, March 24, 2011 12:30 pm, Volker Armin Hemmann wrote: |
| 3 |
> > On Thursday 24 March 2011 08:49:52 J. Roeleveld wrote: |
| 4 |
> >> On Wed, March 23, 2011 5:43 pm, Volker Armin Hemmann wrote: |
| 5 |
> >> > md raid devices can do barriers. Don't know about lvm. But lvm is |
| 6 |
> >> > such |
| 7 |
> >> |
| 8 |
> >> a |
| 9 |
> >> |
| 10 |
> >> > can |
| 11 |
> >> > of worms I am surprised people still recommend it. |
| 12 |
> >> |
| 13 |
> >> What is wrong with LVM? |
| 14 |
> >> I've been using it successfully without any issues for years now. |
| 15 |
> >> It does what it says on the box. |
| 16 |
> > |
| 17 |
> > it is another layer that can go wrong. Why take the risk? There |
| 18 |
> > are enough cases of breakage after upgrades - and besides snapshots... |
| 19 |
> > is |
| 20 |
> > the |
| 21 |
> > amount of additional code running really worth it? Especially with bind |
| 22 |
> > mounting? |
| 23 |
> |
| 24 |
> There are always things that can go wrong and I agree, adding additional |
| 25 |
> layers can increase the risk. |
| 26 |
> However, the benefits of easily and quickly changing the size of |
| 27 |
> partitions and creating snapshots for the use of backups are a big enough |
| 28 |
> benefit to off-set the risk. |
| 29 |
> |
| 30 |
> Bind-mounting is ok, if you use a single filesystem for everything. I have |
| 31 |
> partitions that are filled with thousands of small files and partitions |
| 32 |
> filled with files are are, on average, at 1GB in size. |
| 33 |
> I haven't found a filesystem yet that successfully handles both of these |
| 34 |
> with identical performance. |
| 35 |
> When I first tested performance I found that a simple "ls" in a partition |
| 36 |
> would appear to just hang. This caused performance issues with my |
| 37 |
> IMAP-server. |
| 38 |
> I switched to a filesystem that better handles large amounts of small |
| 39 |
> files and performance increased significantly. |
| 40 |
> |
| 41 |
> The way I do backups is that I stop the services, create snapshots and |
| 42 |
> then restart the services. |
| 43 |
|
| 44 |
and I don't stop much - I just cp everything on the backup disk ;) |
| 45 |
|
| 46 |
/home is way to full? no problem, just dump all that file crap on some other |
| 47 |
partition and bind mount the directories. No change from user POV (that is |
| 48 |
me). And since /home is on / and that a 64gb ssd resizing is not a topic. That |
| 49 |
1tb raid5 I dump all data on? Again - why resize? I have all in one place and |
| 50 |
bind what I need elsewhere. |
Archives