1 |
On Fri, Aug 07, 2015 at 01:39:25PM -0400, Ian Stakenvicius wrote: |
2 |
> -----BEGIN PGP SIGNED MESSAGE----- |
3 |
> Hash: SHA256 |
4 |
> |
5 |
> On 07/08/15 12:59 PM, William Hubbs wrote: |
6 |
> > On Fri, Aug 07, 2015 at 12:10:56PM -0400, Ian Stakenvicius wrote: |
7 |
> >> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 |
8 |
> >> |
9 |
> >> On 07/08/15 11:30 AM, William Hubbs wrote: |
10 |
> >>> On Thu, Aug 06, 2015 at 08:07:44PM -0400, Ian Stakenvicius |
11 |
> >>> wrote: |
12 |
> >>>> |
13 |
> >>>> Can we get "nofail" immediately in the mount -a variants of |
14 |
> >>>> localmount/netmount and expand that in netmount to make the |
15 |
> >>>> nfsclient dep be a "use" or a "need" depending on if it's set |
16 |
> >>>> or not?? That would imo kill the existing bug that started |
17 |
> >>>> all of this too. |
18 |
> >>> |
19 |
> >>> Sure, I can get the nofail support in pretty quick, and I |
20 |
> >>> think that is a feature we should have. |
21 |
> >>> |
22 |
> >>> Right now, netmount is using the use dependency to make sure |
23 |
> >>> network file system utilities are started before us. Because |
24 |
> >>> of the all-or-nothing nature of netmount, we can't switch |
25 |
> >>> those dependencies to need. It would cause netmount to fail if |
26 |
> >>> one of those utilities fails to start. The use dependency is |
27 |
> >>> the best one we can use at this time, and a migration path was |
28 |
> >>> specifically laid out in the news item. |
29 |
> >>> |
30 |
> >> |
31 |
> >> My thinking here is that, unless network mounts in fstab are |
32 |
> >> listed as 'nofail', that netmount failing due to the dependent |
33 |
> >> services not being able to start would be a valid case. |
34 |
> >> Sysadmins that don't want netmount to fail no matter what would |
35 |
> >> be able to use 'nofail' to ensure that happens. |
36 |
> >> |
37 |
> >> This is of course predicated on (1) it being a good idea, and |
38 |
> >> (2) fstabinfo or whatever the check currently is that would add |
39 |
> >> nfsclient to depend() could easily swap 'use' for 'need' based on |
40 |
> >> the (lack of) existence of the nofail attribute in fstab. |
41 |
> > |
42 |
> > The issue with using the need dependency is that netmount is not |
43 |
> > granular enough. It mounts all types of network file systems, so if |
44 |
> > we fail because nfs is a need dependency and doesn't start, no |
45 |
> > other types of network file systems that use daemons will be |
46 |
> > mounted. That's what I meant by all-or-nothing. |
47 |
> > |
48 |
> > William |
49 |
> > |
50 |
> |
51 |
> Yes I follow that. My thoughts are #1, netmount (and localmount for |
52 |
> that matter) are all-or-nothing things; so if the service fails then I |
53 |
> don't think anything can necessarily be assumed as to which mounts |
54 |
> succeeded and which don't. |
55 |
|
56 |
Yes, they are, but mount and umount -a are not. They can report partial |
57 |
failures (check the exit codes). |
58 |
|
59 |
If I switch to using need dependencies for the file system clients, and |
60 |
one fails to start, mount -a would never run, which means NO net-based |
61 |
file systems that use clients would be mounted. |
62 |
|
63 |
My thinking is to allow things to mount that can, but still report it as |
64 |
a failure if everything doesn't mount. |
65 |
|
66 |
Thoughts? |
67 |
|
68 |
William |