build nfsdcld to avoid grace period in containers
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
nfs-utils (Ubuntu) |
Fix Released
|
Undecided
|
Andreas Hasenack |
Bug Description
Trying to use kernel nfs server in containers generally works, but generates dmesg warnings as follows:
[ 23.392559] NFSD: attempt to initialize umh client tracking in a container ignored.
[ 23.395065] NFSD: attempt to initialize legacy client tracking in a container ignored.
[ 23.395085] NFSD: Unable to initialize client recovery tracking! (-22)
[ 23.395106] NFSD: starting 90-second grace period (net f00005fd)
On reboot clients do indeed hang for the grace period.
From the description here ..
https:/
.. it seems that modern kernels in combination with nfsdcld from modern nfs-utils can avoid this problem - is there any chance we could build and install it? It looks like it might have a sqlite dependency so possibly needs splitting off into a separate package.
Related branches
- Bryce Harrington (community): Approve
- Canonical Server: Pending requested
-
Diff: 1586 lines (+1254/-79)12 files modifieddebian/NEWS (+22/-0)
debian/README.Ubuntu (+8/-0)
debian/changelog (+1125/-1)
debian/control (+5/-5)
debian/libnfsidmap1.docs (+1/-0)
debian/libnfsidmap1.install (+3/-1)
debian/nfs-common.docs (+1/-1)
debian/nfs-common.postinst (+4/-0)
debian/not-installed (+5/-0)
debian/patches/remove-regex-from-docs.patch (+79/-0)
debian/patches/series (+1/-0)
dev/null (+0/-71)
description: | updated |
nfs-utils is being updated in bug #1878601 and it does hav nfsdcld