1 |
On Sunday, July 27, 2014 08:44:02 PM Kerin Millar wrote: |
2 |
> On 27/07/2014 17:55, J. Roeleveld wrote: |
3 |
> > On 27 July 2014 18:25:24 CEST, "Stefan G. Weichinger" <lists@×××××.at> |
4 |
wrote: |
5 |
> >> Am 26.07.2014 04:47, schrieb walt: |
6 |
> >>> So, why did the "broken" machine work normally for more than a year |
7 |
> >>> without rpcbind until two days ago? (I suppose because nfs-utils was |
8 |
> >>> updated to 1.3.0 ?) |
9 |
> >>> |
10 |
> >>> The real problem here is that I have no idea how NFS works, and each |
11 |
> >>> new version is more complicated because the devs are solving problems |
12 |
> >>> that I don't understand or even know about. |
13 |
> >> |
14 |
> >> I double your search for understanding ... my various efforts to set up |
15 |
> >> NFSv4 for sharing stuff in my LAN also lead to unstable behavior and |
16 |
> >> frustration. |
17 |
> >> |
18 |
> >> Only last week I re-attacked this topic as I start using puppet here to |
19 |
> >> manage my systems ... and one part of this might be sharing |
20 |
> >> /usr/portage |
21 |
> >> via NFSv4. One client host mounts it without a problem, the thinkpads |
22 |
> >> don't do so ... just another example ;-) |
23 |
> >> |
24 |
> >> Additional in my context: using systemd ... so there are other |
25 |
> >> (different?) dependencies at work and services started. |
26 |
> >> |
27 |
> >> I'd be happy to get that working in a reliable way. I don't remember |
28 |
> >> unstable behavior with NFS (v2 back then?) when we used it at a company |
29 |
> >> I worked for in the 90s. |
30 |
> >> |
31 |
> >> Stefan |
32 |
> > |
33 |
> > I use NFS for filesharing between all wired systems at home. |
34 |
> > Samba is only used for MS Windows and laptops. |
35 |
> > |
36 |
> > Few things I always make sure are valid: |
37 |
> > - One partition per NFS share |
38 |
> > - No NFS share is mounted below another one |
39 |
> > - I set the version to 3 on the clients |
40 |
> > - I use LDAP for the user accounts to ensure the UIDs and GIDs are |
41 |
> > consistent. |
42 |
> These are generally good recommendations. I'd just like to make a few |
43 |
> observations. |
44 |
> |
45 |
> The problems associated with not observing the first constraint (one |
46 |
> filesystem per export) can be alleviated by setting an explicit fsid. |
47 |
> Doing so can also help to avoid stale handles on the client side if the |
48 |
> backing filesystem changes - something that is very useful in a |
49 |
> production environment. Therefore, I tend to start at 1 and increment |
50 |
> with each newly added export. For example:- |
51 |
> |
52 |
> /export/foo *(async,no_subtree_check,fsid=1) |
53 |
> /export/foo/bar *(async,no_subtree_check,fsid=2) |
54 |
> /export/baz *(async,no_subtree_check,fsid=3) |
55 |
> |
56 |
> If using NFSv3, I'd recommend using "nolock" as a mount option unless |
57 |
> there is a genuine requirement for locks to be co-ordinated. Such locks |
58 |
> are only advisory and are of questionable value. Using nolock simplifies |
59 |
> the requirements on both server and client side, and is beneficial for |
60 |
> performance. |
61 |
> |
62 |
> NFSv3/UDP seems to be limited to a maximum read/write block size of |
63 |
> 32768 in Linux, which will be negotiated by default. Using TCP, the |
64 |
> upper bound will be the value of /proc/fs/nfsd/max_block_size on the |
65 |
> server. Its value may be set to 1048576 at the most. NFSv3/TCP is |
66 |
> problematic so I would recommend NFSv4 if TCP is desired as a transport |
67 |
> protocol. |
68 |
> |
69 |
> NFSv4 provides a useful uid/gid mapping feature that is easier to set up |
70 |
> and maintain than nss_ldap. |
71 |
> |
72 |
> > NFS4 requires all the exports to be under a single foldertree. |
73 |
> |
74 |
> This is a myth: |
75 |
> http://linuxcostablanca.blogspot.co.uk/2012/02/nfsv4-myths-and-legends.html. |
76 |
> Exports can be defined and consumed in the same manner as with NFSv3. |
77 |
|
78 |
When I originally tried NFSv4, it refused to work unless they were all under |
79 |
the same directory. |
80 |
As I dislike that, I decided against using it. |
81 |
|
82 |
That was a long time ago, will revisit that part again later. |
83 |
|
84 |
Interesting link, I wonder how difficult it will be to combine that with Samba |
85 |
4 and use the Samba AD structure for NFSv4 with either ZFS or BTRFS |
86 |
underneath. |
87 |
|
88 |
-- |
89 |
Joost |